Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitattributes +4 -0
- grpo_qwen_14B_v2/best_adapter/README.md +207 -0
- grpo_qwen_14B_v2/best_adapter/adapter_config.json +46 -0
- grpo_qwen_14B_v2/best_adapter/adapter_model.safetensors +3 -0
- grpo_qwen_14B_v2/best_adapter/added_tokens.json +24 -0
- grpo_qwen_14B_v2/best_adapter/chat_template.jinja +54 -0
- grpo_qwen_14B_v2/best_adapter/merges.txt +0 -0
- grpo_qwen_14B_v2/best_adapter/special_tokens_map.json +31 -0
- grpo_qwen_14B_v2/best_adapter/tokenizer.json +3 -0
- grpo_qwen_14B_v2/best_adapter/tokenizer_config.json +207 -0
- grpo_qwen_14B_v2/best_adapter/training_args.bin +3 -0
- grpo_qwen_14B_v2/best_adapter/vocab.json +0 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/README.md +207 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/adapter_config.json +46 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/adapter_model.safetensors +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/added_tokens.json +24 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/chat_template.jinja +54 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/merges.txt +0 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/optimizer.pt +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/rng_state.pth +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/scheduler.pt +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/special_tokens_map.json +31 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/tokenizer.json +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/tokenizer_config.json +207 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/trainer_state.json +914 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/training_args.bin +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-400/vocab.json +0 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/README.md +207 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/adapter_config.json +46 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/adapter_model.safetensors +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/added_tokens.json +24 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/chat_template.jinja +54 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/merges.txt +0 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/optimizer.pt +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/rng_state.pth +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/scheduler.pt +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/special_tokens_map.json +31 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/tokenizer.json +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/tokenizer_config.json +207 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/trainer_state.json +1134 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/training_args.bin +3 -0
- grpo_qwen_14B_v2/checkpoints/checkpoint-500/vocab.json +0 -0
- grpo_qwen_14B_v2/config_resolved.yaml +115 -0
- grpo_qwen_14B_v2/logs/grpo_metrics.jsonl +101 -0
- grpo_qwen_14B_v2/logs/train.jsonl +102 -0
- grpo_qwen_14B_v2/wandb/debug-internal.log +12 -0
- grpo_qwen_14B_v2/wandb/debug.log +29 -0
- grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/files/config.yaml +725 -0
- grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/files/output.log +134 -0
- grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/files/requirements.txt +102 -0
.gitattributes
CHANGED
|
@@ -64,3 +64,7 @@ dpo_qwen_14B/checkpoint-100/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
| 64 |
dpo_qwen_14B/wandb/run-20251226_152332-r9hfat2g/run-r9hfat2g.wandb filter=lfs diff=lfs merge=lfs -text
|
| 65 |
dpo_qwen_14B/wandb/run-20251226_152936-r1nptay8/run-r1nptay8.wandb filter=lfs diff=lfs merge=lfs -text
|
| 66 |
dpo_qwen_14B/wandb/run-20251226_155650-wbzoafvt/run-wbzoafvt.wandb filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
dpo_qwen_14B/wandb/run-20251226_152332-r9hfat2g/run-r9hfat2g.wandb filter=lfs diff=lfs merge=lfs -text
|
| 65 |
dpo_qwen_14B/wandb/run-20251226_152936-r1nptay8/run-r1nptay8.wandb filter=lfs diff=lfs merge=lfs -text
|
| 66 |
dpo_qwen_14B/wandb/run-20251226_155650-wbzoafvt/run-wbzoafvt.wandb filter=lfs diff=lfs merge=lfs -text
|
| 67 |
+
grpo_qwen_14B_v2/best_adapter/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 68 |
+
grpo_qwen_14B_v2/checkpoints/checkpoint-400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 69 |
+
grpo_qwen_14B_v2/checkpoints/checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 70 |
+
grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/run-jz7bptqa.wandb filter=lfs diff=lfs merge=lfs -text
|
grpo_qwen_14B_v2/best_adapter/README.md
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 3 |
+
library_name: peft
|
| 4 |
+
pipeline_tag: text-generation
|
| 5 |
+
tags:
|
| 6 |
+
- base_model:adapter:/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 7 |
+
- lora
|
| 8 |
+
- transformers
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Model ID
|
| 12 |
+
|
| 13 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
## Model Details
|
| 18 |
+
|
| 19 |
+
### Model Description
|
| 20 |
+
|
| 21 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
- **Developed by:** [More Information Needed]
|
| 26 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 27 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 28 |
+
- **Model type:** [More Information Needed]
|
| 29 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 30 |
+
- **License:** [More Information Needed]
|
| 31 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 32 |
+
|
| 33 |
+
### Model Sources [optional]
|
| 34 |
+
|
| 35 |
+
<!-- Provide the basic links for the model. -->
|
| 36 |
+
|
| 37 |
+
- **Repository:** [More Information Needed]
|
| 38 |
+
- **Paper [optional]:** [More Information Needed]
|
| 39 |
+
- **Demo [optional]:** [More Information Needed]
|
| 40 |
+
|
| 41 |
+
## Uses
|
| 42 |
+
|
| 43 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 44 |
+
|
| 45 |
+
### Direct Use
|
| 46 |
+
|
| 47 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 48 |
+
|
| 49 |
+
[More Information Needed]
|
| 50 |
+
|
| 51 |
+
### Downstream Use [optional]
|
| 52 |
+
|
| 53 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 54 |
+
|
| 55 |
+
[More Information Needed]
|
| 56 |
+
|
| 57 |
+
### Out-of-Scope Use
|
| 58 |
+
|
| 59 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 60 |
+
|
| 61 |
+
[More Information Needed]
|
| 62 |
+
|
| 63 |
+
## Bias, Risks, and Limitations
|
| 64 |
+
|
| 65 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 66 |
+
|
| 67 |
+
[More Information Needed]
|
| 68 |
+
|
| 69 |
+
### Recommendations
|
| 70 |
+
|
| 71 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 72 |
+
|
| 73 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 74 |
+
|
| 75 |
+
## How to Get Started with the Model
|
| 76 |
+
|
| 77 |
+
Use the code below to get started with the model.
|
| 78 |
+
|
| 79 |
+
[More Information Needed]
|
| 80 |
+
|
| 81 |
+
## Training Details
|
| 82 |
+
|
| 83 |
+
### Training Data
|
| 84 |
+
|
| 85 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 86 |
+
|
| 87 |
+
[More Information Needed]
|
| 88 |
+
|
| 89 |
+
### Training Procedure
|
| 90 |
+
|
| 91 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 92 |
+
|
| 93 |
+
#### Preprocessing [optional]
|
| 94 |
+
|
| 95 |
+
[More Information Needed]
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
#### Training Hyperparameters
|
| 99 |
+
|
| 100 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 101 |
+
|
| 102 |
+
#### Speeds, Sizes, Times [optional]
|
| 103 |
+
|
| 104 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 105 |
+
|
| 106 |
+
[More Information Needed]
|
| 107 |
+
|
| 108 |
+
## Evaluation
|
| 109 |
+
|
| 110 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 111 |
+
|
| 112 |
+
### Testing Data, Factors & Metrics
|
| 113 |
+
|
| 114 |
+
#### Testing Data
|
| 115 |
+
|
| 116 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 117 |
+
|
| 118 |
+
[More Information Needed]
|
| 119 |
+
|
| 120 |
+
#### Factors
|
| 121 |
+
|
| 122 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 123 |
+
|
| 124 |
+
[More Information Needed]
|
| 125 |
+
|
| 126 |
+
#### Metrics
|
| 127 |
+
|
| 128 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 129 |
+
|
| 130 |
+
[More Information Needed]
|
| 131 |
+
|
| 132 |
+
### Results
|
| 133 |
+
|
| 134 |
+
[More Information Needed]
|
| 135 |
+
|
| 136 |
+
#### Summary
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
|
| 140 |
+
## Model Examination [optional]
|
| 141 |
+
|
| 142 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 143 |
+
|
| 144 |
+
[More Information Needed]
|
| 145 |
+
|
| 146 |
+
## Environmental Impact
|
| 147 |
+
|
| 148 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 149 |
+
|
| 150 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 151 |
+
|
| 152 |
+
- **Hardware Type:** [More Information Needed]
|
| 153 |
+
- **Hours used:** [More Information Needed]
|
| 154 |
+
- **Cloud Provider:** [More Information Needed]
|
| 155 |
+
- **Compute Region:** [More Information Needed]
|
| 156 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 157 |
+
|
| 158 |
+
## Technical Specifications [optional]
|
| 159 |
+
|
| 160 |
+
### Model Architecture and Objective
|
| 161 |
+
|
| 162 |
+
[More Information Needed]
|
| 163 |
+
|
| 164 |
+
### Compute Infrastructure
|
| 165 |
+
|
| 166 |
+
[More Information Needed]
|
| 167 |
+
|
| 168 |
+
#### Hardware
|
| 169 |
+
|
| 170 |
+
[More Information Needed]
|
| 171 |
+
|
| 172 |
+
#### Software
|
| 173 |
+
|
| 174 |
+
[More Information Needed]
|
| 175 |
+
|
| 176 |
+
## Citation [optional]
|
| 177 |
+
|
| 178 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 179 |
+
|
| 180 |
+
**BibTeX:**
|
| 181 |
+
|
| 182 |
+
[More Information Needed]
|
| 183 |
+
|
| 184 |
+
**APA:**
|
| 185 |
+
|
| 186 |
+
[More Information Needed]
|
| 187 |
+
|
| 188 |
+
## Glossary [optional]
|
| 189 |
+
|
| 190 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 191 |
+
|
| 192 |
+
[More Information Needed]
|
| 193 |
+
|
| 194 |
+
## More Information [optional]
|
| 195 |
+
|
| 196 |
+
[More Information Needed]
|
| 197 |
+
|
| 198 |
+
## Model Card Authors [optional]
|
| 199 |
+
|
| 200 |
+
[More Information Needed]
|
| 201 |
+
|
| 202 |
+
## Model Card Contact
|
| 203 |
+
|
| 204 |
+
[More Information Needed]
|
| 205 |
+
### Framework versions
|
| 206 |
+
|
| 207 |
+
- PEFT 0.18.0
|
grpo_qwen_14B_v2/best_adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": null,
|
| 6 |
+
"base_model_name_or_path": "/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2",
|
| 7 |
+
"bias": "none",
|
| 8 |
+
"corda_config": null,
|
| 9 |
+
"ensure_weight_tying": false,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 32,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"peft_version": "0.18.0",
|
| 27 |
+
"qalora_group_size": 16,
|
| 28 |
+
"r": 16,
|
| 29 |
+
"rank_pattern": {},
|
| 30 |
+
"revision": null,
|
| 31 |
+
"target_modules": [
|
| 32 |
+
"q_proj",
|
| 33 |
+
"o_proj",
|
| 34 |
+
"v_proj",
|
| 35 |
+
"up_proj",
|
| 36 |
+
"gate_proj",
|
| 37 |
+
"down_proj",
|
| 38 |
+
"k_proj"
|
| 39 |
+
],
|
| 40 |
+
"target_parameters": null,
|
| 41 |
+
"task_type": "CAUSAL_LM",
|
| 42 |
+
"trainable_token_indices": null,
|
| 43 |
+
"use_dora": false,
|
| 44 |
+
"use_qalora": false,
|
| 45 |
+
"use_rslora": false
|
| 46 |
+
}
|
grpo_qwen_14B_v2/best_adapter/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cfddfaff53a19bede615799a30a42fe443ed6813a68f14c0a605ed1e0b75aa7a
|
| 3 |
+
size 275341720
|
grpo_qwen_14B_v2/best_adapter/added_tokens.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"</tool_call>": 151658,
|
| 3 |
+
"<tool_call>": 151657,
|
| 4 |
+
"<|box_end|>": 151649,
|
| 5 |
+
"<|box_start|>": 151648,
|
| 6 |
+
"<|endoftext|>": 151643,
|
| 7 |
+
"<|file_sep|>": 151664,
|
| 8 |
+
"<|fim_middle|>": 151660,
|
| 9 |
+
"<|fim_pad|>": 151662,
|
| 10 |
+
"<|fim_prefix|>": 151659,
|
| 11 |
+
"<|fim_suffix|>": 151661,
|
| 12 |
+
"<|im_end|>": 151645,
|
| 13 |
+
"<|im_start|>": 151644,
|
| 14 |
+
"<|image_pad|>": 151655,
|
| 15 |
+
"<|object_ref_end|>": 151647,
|
| 16 |
+
"<|object_ref_start|>": 151646,
|
| 17 |
+
"<|quad_end|>": 151651,
|
| 18 |
+
"<|quad_start|>": 151650,
|
| 19 |
+
"<|repo_name|>": 151663,
|
| 20 |
+
"<|video_pad|>": 151656,
|
| 21 |
+
"<|vision_end|>": 151653,
|
| 22 |
+
"<|vision_pad|>": 151654,
|
| 23 |
+
"<|vision_start|>": 151652
|
| 24 |
+
}
|
grpo_qwen_14B_v2/best_adapter/chat_template.jinja
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- if tools %}
|
| 2 |
+
{{- '<|im_start|>system\n' }}
|
| 3 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 4 |
+
{{- messages[0]['content'] }}
|
| 5 |
+
{%- else %}
|
| 6 |
+
{{- 'You are a helpful assistant.' }}
|
| 7 |
+
{%- endif %}
|
| 8 |
+
{{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
|
| 9 |
+
{%- for tool in tools %}
|
| 10 |
+
{{- "\n" }}
|
| 11 |
+
{{- tool | tojson }}
|
| 12 |
+
{%- endfor %}
|
| 13 |
+
{{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
|
| 14 |
+
{%- else %}
|
| 15 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 16 |
+
{{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
|
| 17 |
+
{%- else %}
|
| 18 |
+
{{- '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}
|
| 19 |
+
{%- endif %}
|
| 20 |
+
{%- endif %}
|
| 21 |
+
{%- for message in messages %}
|
| 22 |
+
{%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
|
| 23 |
+
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
|
| 24 |
+
{%- elif message.role == "assistant" %}
|
| 25 |
+
{{- '<|im_start|>' + message.role }}
|
| 26 |
+
{%- if message.content %}
|
| 27 |
+
{{- '\n' + message.content }}
|
| 28 |
+
{%- endif %}
|
| 29 |
+
{%- for tool_call in message.tool_calls %}
|
| 30 |
+
{%- if tool_call.function is defined %}
|
| 31 |
+
{%- set tool_call = tool_call.function %}
|
| 32 |
+
{%- endif %}
|
| 33 |
+
{{- '\n<tool_call>\n{"name": "' }}
|
| 34 |
+
{{- tool_call.name }}
|
| 35 |
+
{{- '", "arguments": ' }}
|
| 36 |
+
{{- tool_call.arguments | tojson }}
|
| 37 |
+
{{- '}\n</tool_call>' }}
|
| 38 |
+
{%- endfor %}
|
| 39 |
+
{{- '<|im_end|>\n' }}
|
| 40 |
+
{%- elif message.role == "tool" %}
|
| 41 |
+
{%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
|
| 42 |
+
{{- '<|im_start|>user' }}
|
| 43 |
+
{%- endif %}
|
| 44 |
+
{{- '\n<tool_response>\n' }}
|
| 45 |
+
{{- message.content }}
|
| 46 |
+
{{- '\n</tool_response>' }}
|
| 47 |
+
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
|
| 48 |
+
{{- '<|im_end|>\n' }}
|
| 49 |
+
{%- endif %}
|
| 50 |
+
{%- endif %}
|
| 51 |
+
{%- endfor %}
|
| 52 |
+
{%- if add_generation_prompt %}
|
| 53 |
+
{{- '<|im_start|>assistant\n' }}
|
| 54 |
+
{%- endif %}
|
grpo_qwen_14B_v2/best_adapter/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
grpo_qwen_14B_v2/best_adapter/special_tokens_map.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>",
|
| 5 |
+
"<|object_ref_start|>",
|
| 6 |
+
"<|object_ref_end|>",
|
| 7 |
+
"<|box_start|>",
|
| 8 |
+
"<|box_end|>",
|
| 9 |
+
"<|quad_start|>",
|
| 10 |
+
"<|quad_end|>",
|
| 11 |
+
"<|vision_start|>",
|
| 12 |
+
"<|vision_end|>",
|
| 13 |
+
"<|vision_pad|>",
|
| 14 |
+
"<|image_pad|>",
|
| 15 |
+
"<|video_pad|>"
|
| 16 |
+
],
|
| 17 |
+
"eos_token": {
|
| 18 |
+
"content": "<|endoftext|>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
},
|
| 24 |
+
"pad_token": {
|
| 25 |
+
"content": "<|endoftext|>",
|
| 26 |
+
"lstrip": false,
|
| 27 |
+
"normalized": false,
|
| 28 |
+
"rstrip": false,
|
| 29 |
+
"single_word": false
|
| 30 |
+
}
|
| 31 |
+
}
|
grpo_qwen_14B_v2/best_adapter/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
|
| 3 |
+
size 11421896
|
grpo_qwen_14B_v2/best_adapter/tokenizer_config.json
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"151643": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"151644": {
|
| 14 |
+
"content": "<|im_start|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"151645": {
|
| 22 |
+
"content": "<|im_end|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"151646": {
|
| 30 |
+
"content": "<|object_ref_start|>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"151647": {
|
| 38 |
+
"content": "<|object_ref_end|>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"151648": {
|
| 46 |
+
"content": "<|box_start|>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"151649": {
|
| 54 |
+
"content": "<|box_end|>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
},
|
| 61 |
+
"151650": {
|
| 62 |
+
"content": "<|quad_start|>",
|
| 63 |
+
"lstrip": false,
|
| 64 |
+
"normalized": false,
|
| 65 |
+
"rstrip": false,
|
| 66 |
+
"single_word": false,
|
| 67 |
+
"special": true
|
| 68 |
+
},
|
| 69 |
+
"151651": {
|
| 70 |
+
"content": "<|quad_end|>",
|
| 71 |
+
"lstrip": false,
|
| 72 |
+
"normalized": false,
|
| 73 |
+
"rstrip": false,
|
| 74 |
+
"single_word": false,
|
| 75 |
+
"special": true
|
| 76 |
+
},
|
| 77 |
+
"151652": {
|
| 78 |
+
"content": "<|vision_start|>",
|
| 79 |
+
"lstrip": false,
|
| 80 |
+
"normalized": false,
|
| 81 |
+
"rstrip": false,
|
| 82 |
+
"single_word": false,
|
| 83 |
+
"special": true
|
| 84 |
+
},
|
| 85 |
+
"151653": {
|
| 86 |
+
"content": "<|vision_end|>",
|
| 87 |
+
"lstrip": false,
|
| 88 |
+
"normalized": false,
|
| 89 |
+
"rstrip": false,
|
| 90 |
+
"single_word": false,
|
| 91 |
+
"special": true
|
| 92 |
+
},
|
| 93 |
+
"151654": {
|
| 94 |
+
"content": "<|vision_pad|>",
|
| 95 |
+
"lstrip": false,
|
| 96 |
+
"normalized": false,
|
| 97 |
+
"rstrip": false,
|
| 98 |
+
"single_word": false,
|
| 99 |
+
"special": true
|
| 100 |
+
},
|
| 101 |
+
"151655": {
|
| 102 |
+
"content": "<|image_pad|>",
|
| 103 |
+
"lstrip": false,
|
| 104 |
+
"normalized": false,
|
| 105 |
+
"rstrip": false,
|
| 106 |
+
"single_word": false,
|
| 107 |
+
"special": true
|
| 108 |
+
},
|
| 109 |
+
"151656": {
|
| 110 |
+
"content": "<|video_pad|>",
|
| 111 |
+
"lstrip": false,
|
| 112 |
+
"normalized": false,
|
| 113 |
+
"rstrip": false,
|
| 114 |
+
"single_word": false,
|
| 115 |
+
"special": true
|
| 116 |
+
},
|
| 117 |
+
"151657": {
|
| 118 |
+
"content": "<tool_call>",
|
| 119 |
+
"lstrip": false,
|
| 120 |
+
"normalized": false,
|
| 121 |
+
"rstrip": false,
|
| 122 |
+
"single_word": false,
|
| 123 |
+
"special": false
|
| 124 |
+
},
|
| 125 |
+
"151658": {
|
| 126 |
+
"content": "</tool_call>",
|
| 127 |
+
"lstrip": false,
|
| 128 |
+
"normalized": false,
|
| 129 |
+
"rstrip": false,
|
| 130 |
+
"single_word": false,
|
| 131 |
+
"special": false
|
| 132 |
+
},
|
| 133 |
+
"151659": {
|
| 134 |
+
"content": "<|fim_prefix|>",
|
| 135 |
+
"lstrip": false,
|
| 136 |
+
"normalized": false,
|
| 137 |
+
"rstrip": false,
|
| 138 |
+
"single_word": false,
|
| 139 |
+
"special": false
|
| 140 |
+
},
|
| 141 |
+
"151660": {
|
| 142 |
+
"content": "<|fim_middle|>",
|
| 143 |
+
"lstrip": false,
|
| 144 |
+
"normalized": false,
|
| 145 |
+
"rstrip": false,
|
| 146 |
+
"single_word": false,
|
| 147 |
+
"special": false
|
| 148 |
+
},
|
| 149 |
+
"151661": {
|
| 150 |
+
"content": "<|fim_suffix|>",
|
| 151 |
+
"lstrip": false,
|
| 152 |
+
"normalized": false,
|
| 153 |
+
"rstrip": false,
|
| 154 |
+
"single_word": false,
|
| 155 |
+
"special": false
|
| 156 |
+
},
|
| 157 |
+
"151662": {
|
| 158 |
+
"content": "<|fim_pad|>",
|
| 159 |
+
"lstrip": false,
|
| 160 |
+
"normalized": false,
|
| 161 |
+
"rstrip": false,
|
| 162 |
+
"single_word": false,
|
| 163 |
+
"special": false
|
| 164 |
+
},
|
| 165 |
+
"151663": {
|
| 166 |
+
"content": "<|repo_name|>",
|
| 167 |
+
"lstrip": false,
|
| 168 |
+
"normalized": false,
|
| 169 |
+
"rstrip": false,
|
| 170 |
+
"single_word": false,
|
| 171 |
+
"special": false
|
| 172 |
+
},
|
| 173 |
+
"151664": {
|
| 174 |
+
"content": "<|file_sep|>",
|
| 175 |
+
"lstrip": false,
|
| 176 |
+
"normalized": false,
|
| 177 |
+
"rstrip": false,
|
| 178 |
+
"single_word": false,
|
| 179 |
+
"special": false
|
| 180 |
+
}
|
| 181 |
+
},
|
| 182 |
+
"additional_special_tokens": [
|
| 183 |
+
"<|im_start|>",
|
| 184 |
+
"<|im_end|>",
|
| 185 |
+
"<|object_ref_start|>",
|
| 186 |
+
"<|object_ref_end|>",
|
| 187 |
+
"<|box_start|>",
|
| 188 |
+
"<|box_end|>",
|
| 189 |
+
"<|quad_start|>",
|
| 190 |
+
"<|quad_end|>",
|
| 191 |
+
"<|vision_start|>",
|
| 192 |
+
"<|vision_end|>",
|
| 193 |
+
"<|vision_pad|>",
|
| 194 |
+
"<|image_pad|>",
|
| 195 |
+
"<|video_pad|>"
|
| 196 |
+
],
|
| 197 |
+
"bos_token": null,
|
| 198 |
+
"clean_up_tokenization_spaces": false,
|
| 199 |
+
"eos_token": "<|endoftext|>",
|
| 200 |
+
"errors": "replace",
|
| 201 |
+
"extra_special_tokens": {},
|
| 202 |
+
"model_max_length": 32768,
|
| 203 |
+
"pad_token": "<|endoftext|>",
|
| 204 |
+
"split_special_tokens": false,
|
| 205 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 206 |
+
"unk_token": null
|
| 207 |
+
}
|
grpo_qwen_14B_v2/best_adapter/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a4a4e48ed61b7c96f3bd2836ac828013a311834ab8a9542ea461fe1ff953396b
|
| 3 |
+
size 5496
|
grpo_qwen_14B_v2/best_adapter/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/README.md
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 3 |
+
library_name: peft
|
| 4 |
+
pipeline_tag: text-generation
|
| 5 |
+
tags:
|
| 6 |
+
- base_model:adapter:/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 7 |
+
- lora
|
| 8 |
+
- transformers
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Model ID
|
| 12 |
+
|
| 13 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
## Model Details
|
| 18 |
+
|
| 19 |
+
### Model Description
|
| 20 |
+
|
| 21 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
- **Developed by:** [More Information Needed]
|
| 26 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 27 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 28 |
+
- **Model type:** [More Information Needed]
|
| 29 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 30 |
+
- **License:** [More Information Needed]
|
| 31 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 32 |
+
|
| 33 |
+
### Model Sources [optional]
|
| 34 |
+
|
| 35 |
+
<!-- Provide the basic links for the model. -->
|
| 36 |
+
|
| 37 |
+
- **Repository:** [More Information Needed]
|
| 38 |
+
- **Paper [optional]:** [More Information Needed]
|
| 39 |
+
- **Demo [optional]:** [More Information Needed]
|
| 40 |
+
|
| 41 |
+
## Uses
|
| 42 |
+
|
| 43 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 44 |
+
|
| 45 |
+
### Direct Use
|
| 46 |
+
|
| 47 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 48 |
+
|
| 49 |
+
[More Information Needed]
|
| 50 |
+
|
| 51 |
+
### Downstream Use [optional]
|
| 52 |
+
|
| 53 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 54 |
+
|
| 55 |
+
[More Information Needed]
|
| 56 |
+
|
| 57 |
+
### Out-of-Scope Use
|
| 58 |
+
|
| 59 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 60 |
+
|
| 61 |
+
[More Information Needed]
|
| 62 |
+
|
| 63 |
+
## Bias, Risks, and Limitations
|
| 64 |
+
|
| 65 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 66 |
+
|
| 67 |
+
[More Information Needed]
|
| 68 |
+
|
| 69 |
+
### Recommendations
|
| 70 |
+
|
| 71 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 72 |
+
|
| 73 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 74 |
+
|
| 75 |
+
## How to Get Started with the Model
|
| 76 |
+
|
| 77 |
+
Use the code below to get started with the model.
|
| 78 |
+
|
| 79 |
+
[More Information Needed]
|
| 80 |
+
|
| 81 |
+
## Training Details
|
| 82 |
+
|
| 83 |
+
### Training Data
|
| 84 |
+
|
| 85 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 86 |
+
|
| 87 |
+
[More Information Needed]
|
| 88 |
+
|
| 89 |
+
### Training Procedure
|
| 90 |
+
|
| 91 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 92 |
+
|
| 93 |
+
#### Preprocessing [optional]
|
| 94 |
+
|
| 95 |
+
[More Information Needed]
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
#### Training Hyperparameters
|
| 99 |
+
|
| 100 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 101 |
+
|
| 102 |
+
#### Speeds, Sizes, Times [optional]
|
| 103 |
+
|
| 104 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 105 |
+
|
| 106 |
+
[More Information Needed]
|
| 107 |
+
|
| 108 |
+
## Evaluation
|
| 109 |
+
|
| 110 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 111 |
+
|
| 112 |
+
### Testing Data, Factors & Metrics
|
| 113 |
+
|
| 114 |
+
#### Testing Data
|
| 115 |
+
|
| 116 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 117 |
+
|
| 118 |
+
[More Information Needed]
|
| 119 |
+
|
| 120 |
+
#### Factors
|
| 121 |
+
|
| 122 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 123 |
+
|
| 124 |
+
[More Information Needed]
|
| 125 |
+
|
| 126 |
+
#### Metrics
|
| 127 |
+
|
| 128 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 129 |
+
|
| 130 |
+
[More Information Needed]
|
| 131 |
+
|
| 132 |
+
### Results
|
| 133 |
+
|
| 134 |
+
[More Information Needed]
|
| 135 |
+
|
| 136 |
+
#### Summary
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
|
| 140 |
+
## Model Examination [optional]
|
| 141 |
+
|
| 142 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 143 |
+
|
| 144 |
+
[More Information Needed]
|
| 145 |
+
|
| 146 |
+
## Environmental Impact
|
| 147 |
+
|
| 148 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 149 |
+
|
| 150 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 151 |
+
|
| 152 |
+
- **Hardware Type:** [More Information Needed]
|
| 153 |
+
- **Hours used:** [More Information Needed]
|
| 154 |
+
- **Cloud Provider:** [More Information Needed]
|
| 155 |
+
- **Compute Region:** [More Information Needed]
|
| 156 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 157 |
+
|
| 158 |
+
## Technical Specifications [optional]
|
| 159 |
+
|
| 160 |
+
### Model Architecture and Objective
|
| 161 |
+
|
| 162 |
+
[More Information Needed]
|
| 163 |
+
|
| 164 |
+
### Compute Infrastructure
|
| 165 |
+
|
| 166 |
+
[More Information Needed]
|
| 167 |
+
|
| 168 |
+
#### Hardware
|
| 169 |
+
|
| 170 |
+
[More Information Needed]
|
| 171 |
+
|
| 172 |
+
#### Software
|
| 173 |
+
|
| 174 |
+
[More Information Needed]
|
| 175 |
+
|
| 176 |
+
## Citation [optional]
|
| 177 |
+
|
| 178 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 179 |
+
|
| 180 |
+
**BibTeX:**
|
| 181 |
+
|
| 182 |
+
[More Information Needed]
|
| 183 |
+
|
| 184 |
+
**APA:**
|
| 185 |
+
|
| 186 |
+
[More Information Needed]
|
| 187 |
+
|
| 188 |
+
## Glossary [optional]
|
| 189 |
+
|
| 190 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 191 |
+
|
| 192 |
+
[More Information Needed]
|
| 193 |
+
|
| 194 |
+
## More Information [optional]
|
| 195 |
+
|
| 196 |
+
[More Information Needed]
|
| 197 |
+
|
| 198 |
+
## Model Card Authors [optional]
|
| 199 |
+
|
| 200 |
+
[More Information Needed]
|
| 201 |
+
|
| 202 |
+
## Model Card Contact
|
| 203 |
+
|
| 204 |
+
[More Information Needed]
|
| 205 |
+
### Framework versions
|
| 206 |
+
|
| 207 |
+
- PEFT 0.18.0
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/adapter_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": null,
|
| 6 |
+
"base_model_name_or_path": "/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2",
|
| 7 |
+
"bias": "none",
|
| 8 |
+
"corda_config": null,
|
| 9 |
+
"ensure_weight_tying": false,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 32,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"peft_version": "0.18.0",
|
| 27 |
+
"qalora_group_size": 16,
|
| 28 |
+
"r": 16,
|
| 29 |
+
"rank_pattern": {},
|
| 30 |
+
"revision": null,
|
| 31 |
+
"target_modules": [
|
| 32 |
+
"q_proj",
|
| 33 |
+
"o_proj",
|
| 34 |
+
"v_proj",
|
| 35 |
+
"up_proj",
|
| 36 |
+
"gate_proj",
|
| 37 |
+
"down_proj",
|
| 38 |
+
"k_proj"
|
| 39 |
+
],
|
| 40 |
+
"target_parameters": null,
|
| 41 |
+
"task_type": "CAUSAL_LM",
|
| 42 |
+
"trainable_token_indices": null,
|
| 43 |
+
"use_dora": false,
|
| 44 |
+
"use_qalora": false,
|
| 45 |
+
"use_rslora": false
|
| 46 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a30d78957cdbfa18908fab85c7673f6fc53d959c42542fa7a0679c258a0bc12
|
| 3 |
+
size 275341720
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/added_tokens.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"</tool_call>": 151658,
|
| 3 |
+
"<tool_call>": 151657,
|
| 4 |
+
"<|box_end|>": 151649,
|
| 5 |
+
"<|box_start|>": 151648,
|
| 6 |
+
"<|endoftext|>": 151643,
|
| 7 |
+
"<|file_sep|>": 151664,
|
| 8 |
+
"<|fim_middle|>": 151660,
|
| 9 |
+
"<|fim_pad|>": 151662,
|
| 10 |
+
"<|fim_prefix|>": 151659,
|
| 11 |
+
"<|fim_suffix|>": 151661,
|
| 12 |
+
"<|im_end|>": 151645,
|
| 13 |
+
"<|im_start|>": 151644,
|
| 14 |
+
"<|image_pad|>": 151655,
|
| 15 |
+
"<|object_ref_end|>": 151647,
|
| 16 |
+
"<|object_ref_start|>": 151646,
|
| 17 |
+
"<|quad_end|>": 151651,
|
| 18 |
+
"<|quad_start|>": 151650,
|
| 19 |
+
"<|repo_name|>": 151663,
|
| 20 |
+
"<|video_pad|>": 151656,
|
| 21 |
+
"<|vision_end|>": 151653,
|
| 22 |
+
"<|vision_pad|>": 151654,
|
| 23 |
+
"<|vision_start|>": 151652
|
| 24 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/chat_template.jinja
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- if tools %}
|
| 2 |
+
{{- '<|im_start|>system\n' }}
|
| 3 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 4 |
+
{{- messages[0]['content'] }}
|
| 5 |
+
{%- else %}
|
| 6 |
+
{{- 'You are a helpful assistant.' }}
|
| 7 |
+
{%- endif %}
|
| 8 |
+
{{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
|
| 9 |
+
{%- for tool in tools %}
|
| 10 |
+
{{- "\n" }}
|
| 11 |
+
{{- tool | tojson }}
|
| 12 |
+
{%- endfor %}
|
| 13 |
+
{{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
|
| 14 |
+
{%- else %}
|
| 15 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 16 |
+
{{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
|
| 17 |
+
{%- else %}
|
| 18 |
+
{{- '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}
|
| 19 |
+
{%- endif %}
|
| 20 |
+
{%- endif %}
|
| 21 |
+
{%- for message in messages %}
|
| 22 |
+
{%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
|
| 23 |
+
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
|
| 24 |
+
{%- elif message.role == "assistant" %}
|
| 25 |
+
{{- '<|im_start|>' + message.role }}
|
| 26 |
+
{%- if message.content %}
|
| 27 |
+
{{- '\n' + message.content }}
|
| 28 |
+
{%- endif %}
|
| 29 |
+
{%- for tool_call in message.tool_calls %}
|
| 30 |
+
{%- if tool_call.function is defined %}
|
| 31 |
+
{%- set tool_call = tool_call.function %}
|
| 32 |
+
{%- endif %}
|
| 33 |
+
{{- '\n<tool_call>\n{"name": "' }}
|
| 34 |
+
{{- tool_call.name }}
|
| 35 |
+
{{- '", "arguments": ' }}
|
| 36 |
+
{{- tool_call.arguments | tojson }}
|
| 37 |
+
{{- '}\n</tool_call>' }}
|
| 38 |
+
{%- endfor %}
|
| 39 |
+
{{- '<|im_end|>\n' }}
|
| 40 |
+
{%- elif message.role == "tool" %}
|
| 41 |
+
{%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
|
| 42 |
+
{{- '<|im_start|>user' }}
|
| 43 |
+
{%- endif %}
|
| 44 |
+
{{- '\n<tool_response>\n' }}
|
| 45 |
+
{{- message.content }}
|
| 46 |
+
{{- '\n</tool_response>' }}
|
| 47 |
+
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
|
| 48 |
+
{{- '<|im_end|>\n' }}
|
| 49 |
+
{%- endif %}
|
| 50 |
+
{%- endif %}
|
| 51 |
+
{%- endfor %}
|
| 52 |
+
{%- if add_generation_prompt %}
|
| 53 |
+
{{- '<|im_start|>assistant\n' }}
|
| 54 |
+
{%- endif %}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:061d403dcb81ead17fa966d89635c2c9386b81af84fa79f38fe72b302078ce56
|
| 3 |
+
size 551070514
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df773721583080a5937ae757c6cabb41d190e187d9dcba072940a8fb11a22714
|
| 3 |
+
size 14244
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f06565d55295681785448bb88140fb9d768cda54cb2f3e5bfc6f0d216a81cfc8
|
| 3 |
+
size 1064
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/special_tokens_map.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>",
|
| 5 |
+
"<|object_ref_start|>",
|
| 6 |
+
"<|object_ref_end|>",
|
| 7 |
+
"<|box_start|>",
|
| 8 |
+
"<|box_end|>",
|
| 9 |
+
"<|quad_start|>",
|
| 10 |
+
"<|quad_end|>",
|
| 11 |
+
"<|vision_start|>",
|
| 12 |
+
"<|vision_end|>",
|
| 13 |
+
"<|vision_pad|>",
|
| 14 |
+
"<|image_pad|>",
|
| 15 |
+
"<|video_pad|>"
|
| 16 |
+
],
|
| 17 |
+
"eos_token": {
|
| 18 |
+
"content": "<|endoftext|>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
},
|
| 24 |
+
"pad_token": {
|
| 25 |
+
"content": "<|endoftext|>",
|
| 26 |
+
"lstrip": false,
|
| 27 |
+
"normalized": false,
|
| 28 |
+
"rstrip": false,
|
| 29 |
+
"single_word": false
|
| 30 |
+
}
|
| 31 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
|
| 3 |
+
size 11421896
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/tokenizer_config.json
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"151643": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"151644": {
|
| 14 |
+
"content": "<|im_start|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"151645": {
|
| 22 |
+
"content": "<|im_end|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"151646": {
|
| 30 |
+
"content": "<|object_ref_start|>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"151647": {
|
| 38 |
+
"content": "<|object_ref_end|>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"151648": {
|
| 46 |
+
"content": "<|box_start|>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"151649": {
|
| 54 |
+
"content": "<|box_end|>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
},
|
| 61 |
+
"151650": {
|
| 62 |
+
"content": "<|quad_start|>",
|
| 63 |
+
"lstrip": false,
|
| 64 |
+
"normalized": false,
|
| 65 |
+
"rstrip": false,
|
| 66 |
+
"single_word": false,
|
| 67 |
+
"special": true
|
| 68 |
+
},
|
| 69 |
+
"151651": {
|
| 70 |
+
"content": "<|quad_end|>",
|
| 71 |
+
"lstrip": false,
|
| 72 |
+
"normalized": false,
|
| 73 |
+
"rstrip": false,
|
| 74 |
+
"single_word": false,
|
| 75 |
+
"special": true
|
| 76 |
+
},
|
| 77 |
+
"151652": {
|
| 78 |
+
"content": "<|vision_start|>",
|
| 79 |
+
"lstrip": false,
|
| 80 |
+
"normalized": false,
|
| 81 |
+
"rstrip": false,
|
| 82 |
+
"single_word": false,
|
| 83 |
+
"special": true
|
| 84 |
+
},
|
| 85 |
+
"151653": {
|
| 86 |
+
"content": "<|vision_end|>",
|
| 87 |
+
"lstrip": false,
|
| 88 |
+
"normalized": false,
|
| 89 |
+
"rstrip": false,
|
| 90 |
+
"single_word": false,
|
| 91 |
+
"special": true
|
| 92 |
+
},
|
| 93 |
+
"151654": {
|
| 94 |
+
"content": "<|vision_pad|>",
|
| 95 |
+
"lstrip": false,
|
| 96 |
+
"normalized": false,
|
| 97 |
+
"rstrip": false,
|
| 98 |
+
"single_word": false,
|
| 99 |
+
"special": true
|
| 100 |
+
},
|
| 101 |
+
"151655": {
|
| 102 |
+
"content": "<|image_pad|>",
|
| 103 |
+
"lstrip": false,
|
| 104 |
+
"normalized": false,
|
| 105 |
+
"rstrip": false,
|
| 106 |
+
"single_word": false,
|
| 107 |
+
"special": true
|
| 108 |
+
},
|
| 109 |
+
"151656": {
|
| 110 |
+
"content": "<|video_pad|>",
|
| 111 |
+
"lstrip": false,
|
| 112 |
+
"normalized": false,
|
| 113 |
+
"rstrip": false,
|
| 114 |
+
"single_word": false,
|
| 115 |
+
"special": true
|
| 116 |
+
},
|
| 117 |
+
"151657": {
|
| 118 |
+
"content": "<tool_call>",
|
| 119 |
+
"lstrip": false,
|
| 120 |
+
"normalized": false,
|
| 121 |
+
"rstrip": false,
|
| 122 |
+
"single_word": false,
|
| 123 |
+
"special": false
|
| 124 |
+
},
|
| 125 |
+
"151658": {
|
| 126 |
+
"content": "</tool_call>",
|
| 127 |
+
"lstrip": false,
|
| 128 |
+
"normalized": false,
|
| 129 |
+
"rstrip": false,
|
| 130 |
+
"single_word": false,
|
| 131 |
+
"special": false
|
| 132 |
+
},
|
| 133 |
+
"151659": {
|
| 134 |
+
"content": "<|fim_prefix|>",
|
| 135 |
+
"lstrip": false,
|
| 136 |
+
"normalized": false,
|
| 137 |
+
"rstrip": false,
|
| 138 |
+
"single_word": false,
|
| 139 |
+
"special": false
|
| 140 |
+
},
|
| 141 |
+
"151660": {
|
| 142 |
+
"content": "<|fim_middle|>",
|
| 143 |
+
"lstrip": false,
|
| 144 |
+
"normalized": false,
|
| 145 |
+
"rstrip": false,
|
| 146 |
+
"single_word": false,
|
| 147 |
+
"special": false
|
| 148 |
+
},
|
| 149 |
+
"151661": {
|
| 150 |
+
"content": "<|fim_suffix|>",
|
| 151 |
+
"lstrip": false,
|
| 152 |
+
"normalized": false,
|
| 153 |
+
"rstrip": false,
|
| 154 |
+
"single_word": false,
|
| 155 |
+
"special": false
|
| 156 |
+
},
|
| 157 |
+
"151662": {
|
| 158 |
+
"content": "<|fim_pad|>",
|
| 159 |
+
"lstrip": false,
|
| 160 |
+
"normalized": false,
|
| 161 |
+
"rstrip": false,
|
| 162 |
+
"single_word": false,
|
| 163 |
+
"special": false
|
| 164 |
+
},
|
| 165 |
+
"151663": {
|
| 166 |
+
"content": "<|repo_name|>",
|
| 167 |
+
"lstrip": false,
|
| 168 |
+
"normalized": false,
|
| 169 |
+
"rstrip": false,
|
| 170 |
+
"single_word": false,
|
| 171 |
+
"special": false
|
| 172 |
+
},
|
| 173 |
+
"151664": {
|
| 174 |
+
"content": "<|file_sep|>",
|
| 175 |
+
"lstrip": false,
|
| 176 |
+
"normalized": false,
|
| 177 |
+
"rstrip": false,
|
| 178 |
+
"single_word": false,
|
| 179 |
+
"special": false
|
| 180 |
+
}
|
| 181 |
+
},
|
| 182 |
+
"additional_special_tokens": [
|
| 183 |
+
"<|im_start|>",
|
| 184 |
+
"<|im_end|>",
|
| 185 |
+
"<|object_ref_start|>",
|
| 186 |
+
"<|object_ref_end|>",
|
| 187 |
+
"<|box_start|>",
|
| 188 |
+
"<|box_end|>",
|
| 189 |
+
"<|quad_start|>",
|
| 190 |
+
"<|quad_end|>",
|
| 191 |
+
"<|vision_start|>",
|
| 192 |
+
"<|vision_end|>",
|
| 193 |
+
"<|vision_pad|>",
|
| 194 |
+
"<|image_pad|>",
|
| 195 |
+
"<|video_pad|>"
|
| 196 |
+
],
|
| 197 |
+
"bos_token": null,
|
| 198 |
+
"clean_up_tokenization_spaces": false,
|
| 199 |
+
"eos_token": "<|endoftext|>",
|
| 200 |
+
"errors": "replace",
|
| 201 |
+
"extra_special_tokens": {},
|
| 202 |
+
"model_max_length": 32768,
|
| 203 |
+
"pad_token": "<|endoftext|>",
|
| 204 |
+
"split_special_tokens": false,
|
| 205 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 206 |
+
"unk_token": null
|
| 207 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/trainer_state.json
ADDED
|
@@ -0,0 +1,914 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.6006006006006006,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 400,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02002002002002002,
|
| 14 |
+
"grad_norm": 0.05460292845964432,
|
| 15 |
+
"grpo_mean_advantage": -1.3560057254835556e-07,
|
| 16 |
+
"grpo_mean_group_score": 0.5922331809997559,
|
| 17 |
+
"grpo_mean_kl_div": 0.0,
|
| 18 |
+
"grpo_std_advantage": 3.0318567496578908e-06,
|
| 19 |
+
"learning_rate": 8.000000000000001e-07,
|
| 20 |
+
"loss": 0.007,
|
| 21 |
+
"step": 5
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.04004004004004004,
|
| 25 |
+
"grad_norm": 0.0679207444190979,
|
| 26 |
+
"grpo_mean_advantage": 3.6619603633880615e-06,
|
| 27 |
+
"grpo_mean_group_score": 0.5561589002609253,
|
| 28 |
+
"grpo_mean_kl_div": 0.0,
|
| 29 |
+
"grpo_std_advantage": 1.6246918676188216e-05,
|
| 30 |
+
"learning_rate": 1.8000000000000001e-06,
|
| 31 |
+
"loss": 0.0107,
|
| 32 |
+
"step": 10
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06006006006006006,
|
| 36 |
+
"grad_norm": 0.05788416787981987,
|
| 37 |
+
"grpo_mean_advantage": -1.0654330395709621e-07,
|
| 38 |
+
"grpo_mean_group_score": 0.5759152173995972,
|
| 39 |
+
"grpo_mean_kl_div": 0.0,
|
| 40 |
+
"grpo_std_advantage": 5.399440965447866e-07,
|
| 41 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 42 |
+
"loss": 0.007,
|
| 43 |
+
"step": 15
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.08008008008008008,
|
| 47 |
+
"grad_norm": 0.0746568813920021,
|
| 48 |
+
"grpo_mean_advantage": -5.871057737749652e-07,
|
| 49 |
+
"grpo_mean_group_score": 0.5127314329147339,
|
| 50 |
+
"grpo_mean_kl_div": 0.0,
|
| 51 |
+
"grpo_std_advantage": 2.6951597646984737e-06,
|
| 52 |
+
"learning_rate": 3.8000000000000005e-06,
|
| 53 |
+
"loss": 0.0246,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.1001001001001001,
|
| 58 |
+
"grad_norm": 0.11442846059799194,
|
| 59 |
+
"grpo_mean_advantage": 6.370246410369873e-07,
|
| 60 |
+
"grpo_mean_group_score": 0.539706826210022,
|
| 61 |
+
"grpo_mean_kl_div": 0.0,
|
| 62 |
+
"grpo_std_advantage": 2.8908377771585947e-06,
|
| 63 |
+
"learning_rate": 4.800000000000001e-06,
|
| 64 |
+
"loss": 0.0337,
|
| 65 |
+
"step": 25
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.12012012012012012,
|
| 69 |
+
"grad_norm": 0.05778791010379791,
|
| 70 |
+
"grpo_mean_advantage": 6.705522359595761e-09,
|
| 71 |
+
"grpo_mean_group_score": 0.5812538862228394,
|
| 72 |
+
"grpo_mean_kl_div": 0.0,
|
| 73 |
+
"grpo_std_advantage": 6.189450800775376e-07,
|
| 74 |
+
"learning_rate": 4.999125183044924e-06,
|
| 75 |
+
"loss": 0.0171,
|
| 76 |
+
"step": 30
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.14014014014014015,
|
| 80 |
+
"grad_norm": 0.05819695070385933,
|
| 81 |
+
"grpo_mean_advantage": 3.859400692363124e-07,
|
| 82 |
+
"grpo_mean_group_score": 0.5909844636917114,
|
| 83 |
+
"grpo_mean_kl_div": 0.0,
|
| 84 |
+
"grpo_std_advantage": 1.6833292875162442e-06,
|
| 85 |
+
"learning_rate": 4.995572288443412e-06,
|
| 86 |
+
"loss": 0.0145,
|
| 87 |
+
"step": 35
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.16016016016016016,
|
| 91 |
+
"grad_norm": 0.07968433201313019,
|
| 92 |
+
"grpo_mean_advantage": 2.600252742013254e-07,
|
| 93 |
+
"grpo_mean_group_score": 0.5630953907966614,
|
| 94 |
+
"grpo_mean_kl_div": 0.0,
|
| 95 |
+
"grpo_std_advantage": 1.4095899132371414e-06,
|
| 96 |
+
"learning_rate": 4.98929052218411e-06,
|
| 97 |
+
"loss": 0.0196,
|
| 98 |
+
"step": 40
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.18018018018018017,
|
| 102 |
+
"grad_norm": 0.0733402892947197,
|
| 103 |
+
"grpo_mean_advantage": -1.2591480924584175e-07,
|
| 104 |
+
"grpo_mean_group_score": 0.5604403614997864,
|
| 105 |
+
"grpo_mean_kl_div": 0.0,
|
| 106 |
+
"grpo_std_advantage": 1.0309080380466185e-06,
|
| 107 |
+
"learning_rate": 4.980286753286196e-06,
|
| 108 |
+
"loss": 0.0186,
|
| 109 |
+
"step": 45
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.2002002002002002,
|
| 113 |
+
"grad_norm": 0.07136482000350952,
|
| 114 |
+
"grpo_mean_advantage": -2.808868941883702e-07,
|
| 115 |
+
"grpo_mean_group_score": 0.5971035957336426,
|
| 116 |
+
"grpo_mean_kl_div": 0.0,
|
| 117 |
+
"grpo_std_advantage": 1.5696078889959608e-06,
|
| 118 |
+
"learning_rate": 4.9685708272387645e-06,
|
| 119 |
+
"loss": 0.0286,
|
| 120 |
+
"step": 50
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.22022022022022023,
|
| 124 |
+
"grad_norm": 0.08851475268602371,
|
| 125 |
+
"grpo_mean_advantage": 2.6822089438383045e-08,
|
| 126 |
+
"grpo_mean_group_score": 0.5892971754074097,
|
| 127 |
+
"grpo_mean_kl_div": 0.0,
|
| 128 |
+
"grpo_std_advantage": 3.7878271541558206e-07,
|
| 129 |
+
"learning_rate": 4.9541555552349404e-06,
|
| 130 |
+
"loss": 0.0054,
|
| 131 |
+
"step": 55
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.24024024024024024,
|
| 135 |
+
"grad_norm": 0.07778509706258774,
|
| 136 |
+
"grpo_mean_advantage": -5.662441182607836e-08,
|
| 137 |
+
"grpo_mean_group_score": 0.564322292804718,
|
| 138 |
+
"grpo_mean_kl_div": 0.0,
|
| 139 |
+
"grpo_std_advantage": 6.128998393251095e-07,
|
| 140 |
+
"learning_rate": 4.9370567001630155e-06,
|
| 141 |
+
"loss": -0.0074,
|
| 142 |
+
"step": 60
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2602602602602603,
|
| 146 |
+
"grad_norm": 0.08740051090717316,
|
| 147 |
+
"grpo_mean_advantage": -1.5944242193199898e-07,
|
| 148 |
+
"grpo_mean_group_score": 0.562497615814209,
|
| 149 |
+
"grpo_mean_kl_div": 0.0,
|
| 150 |
+
"grpo_std_advantage": 1.6374274309782777e-06,
|
| 151 |
+
"learning_rate": 4.917292959369968e-06,
|
| 152 |
+
"loss": 0.0145,
|
| 153 |
+
"step": 65
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.2802802802802803,
|
| 157 |
+
"grad_norm": 0.19070060551166534,
|
| 158 |
+
"grpo_mean_advantage": 1.6838312433264946e-07,
|
| 159 |
+
"grpo_mean_group_score": 0.5904761552810669,
|
| 160 |
+
"grpo_mean_kl_div": 0.0,
|
| 161 |
+
"grpo_std_advantage": 8.536571272088622e-07,
|
| 162 |
+
"learning_rate": 4.8948859442161876e-06,
|
| 163 |
+
"loss": 0.0257,
|
| 164 |
+
"step": 70
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3003003003003003,
|
| 168 |
+
"grad_norm": 0.07321271300315857,
|
| 169 |
+
"grpo_mean_advantage": 1.1175870895385742e-07,
|
| 170 |
+
"grpo_mean_group_score": 0.5765624046325684,
|
| 171 |
+
"grpo_mean_kl_div": 0.0,
|
| 172 |
+
"grpo_std_advantage": 6.451961667153228e-07,
|
| 173 |
+
"learning_rate": 4.869860156443768e-06,
|
| 174 |
+
"loss": 0.0024,
|
| 175 |
+
"step": 75
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.3203203203203203,
|
| 179 |
+
"grad_norm": 0.07126748561859131,
|
| 180 |
+
"grpo_mean_advantage": -1.4603138254187797e-07,
|
| 181 |
+
"grpo_mean_group_score": 0.5858271718025208,
|
| 182 |
+
"grpo_mean_kl_div": 0.0,
|
| 183 |
+
"grpo_std_advantage": 1.1309343790344428e-06,
|
| 184 |
+
"learning_rate": 4.842242961384211e-06,
|
| 185 |
+
"loss": 0.0277,
|
| 186 |
+
"step": 80
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.34034034034034033,
|
| 190 |
+
"grad_norm": 0.08629189431667328,
|
| 191 |
+
"grpo_mean_advantage": -1.817941665649414e-06,
|
| 192 |
+
"grpo_mean_group_score": 0.5871662497520447,
|
| 193 |
+
"grpo_mean_kl_div": 0.0,
|
| 194 |
+
"grpo_std_advantage": 1.1141768482048064e-05,
|
| 195 |
+
"learning_rate": 4.812064558034847e-06,
|
| 196 |
+
"loss": 0.0246,
|
| 197 |
+
"step": 85
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.36036036036036034,
|
| 201 |
+
"grad_norm": 0.0998779758810997,
|
| 202 |
+
"grpo_mean_advantage": 1.8179416372277046e-07,
|
| 203 |
+
"grpo_mean_group_score": 0.5330992937088013,
|
| 204 |
+
"grpo_mean_kl_div": 0.0,
|
| 205 |
+
"grpo_std_advantage": 6.210335072864837e-07,
|
| 206 |
+
"learning_rate": 4.779357946036662e-06,
|
| 207 |
+
"loss": 0.0056,
|
| 208 |
+
"step": 90
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.38038038038038036,
|
| 212 |
+
"grad_norm": 0.10614689439535141,
|
| 213 |
+
"grpo_mean_advantage": -2.972781771859445e-07,
|
| 214 |
+
"grpo_mean_group_score": 0.5265295505523682,
|
| 215 |
+
"grpo_mean_kl_div": 0.0,
|
| 216 |
+
"grpo_std_advantage": 3.1582342217006953e-06,
|
| 217 |
+
"learning_rate": 4.74415888958968e-06,
|
| 218 |
+
"loss": 0.0053,
|
| 219 |
+
"step": 95
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.4004004004004004,
|
| 223 |
+
"grad_norm": 0.10345634073019028,
|
| 224 |
+
"grpo_mean_advantage": -7.033348197182931e-07,
|
| 225 |
+
"grpo_mean_group_score": 0.5660771131515503,
|
| 226 |
+
"grpo_mean_kl_div": 0.0,
|
| 227 |
+
"grpo_std_advantage": 4.245831405569334e-06,
|
| 228 |
+
"learning_rate": 4.706505878345343e-06,
|
| 229 |
+
"loss": 0.0134,
|
| 230 |
+
"step": 100
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.42042042042042044,
|
| 234 |
+
"grad_norm": 0.10077933222055435,
|
| 235 |
+
"grpo_mean_advantage": 1.1920928955078125e-07,
|
| 236 |
+
"grpo_mean_group_score": 0.57631915807724,
|
| 237 |
+
"grpo_mean_kl_div": 0.0,
|
| 238 |
+
"grpo_std_advantage": 3.2809634831210133e-07,
|
| 239 |
+
"learning_rate": 4.666440085318626e-06,
|
| 240 |
+
"loss": 0.0004,
|
| 241 |
+
"step": 105
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.44044044044044045,
|
| 245 |
+
"grad_norm": 0.09548182785511017,
|
| 246 |
+
"grpo_mean_advantage": -4.0978193283081055e-07,
|
| 247 |
+
"grpo_mean_group_score": 0.546563982963562,
|
| 248 |
+
"grpo_mean_kl_div": 0.0,
|
| 249 |
+
"grpo_std_advantage": 6.0397578636184335e-06,
|
| 250 |
+
"learning_rate": 4.624005321865968e-06,
|
| 251 |
+
"loss": 0.0033,
|
| 252 |
+
"step": 110
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.46046046046046046,
|
| 256 |
+
"grad_norm": 0.09417816251516342,
|
| 257 |
+
"grpo_mean_advantage": -1.467764434437413e-07,
|
| 258 |
+
"grpo_mean_group_score": 0.5519219636917114,
|
| 259 |
+
"grpo_mean_kl_div": 0.0,
|
| 260 |
+
"grpo_std_advantage": 2.2689375782647403e-06,
|
| 261 |
+
"learning_rate": 4.57924798977818e-06,
|
| 262 |
+
"loss": 0.0095,
|
| 263 |
+
"step": 115
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.4804804804804805,
|
| 267 |
+
"grad_norm": 0.10022275149822235,
|
| 268 |
+
"grpo_mean_advantage": -5.215406329028838e-09,
|
| 269 |
+
"grpo_mean_group_score": 0.5490407943725586,
|
| 270 |
+
"grpo_mean_kl_div": 0.0,
|
| 271 |
+
"grpo_std_advantage": 7.929010621410271e-07,
|
| 272 |
+
"learning_rate": 4.532217030540781e-06,
|
| 273 |
+
"loss": 0.0006,
|
| 274 |
+
"step": 120
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.5005005005005005,
|
| 278 |
+
"grad_norm": 0.14057794213294983,
|
| 279 |
+
"grpo_mean_advantage": -5.7369469175228005e-08,
|
| 280 |
+
"grpo_mean_group_score": 0.5646580457687378,
|
| 281 |
+
"grpo_mean_kl_div": 0.0,
|
| 282 |
+
"grpo_std_advantage": 1.2823379620385822e-06,
|
| 283 |
+
"learning_rate": 4.482963871817195e-06,
|
| 284 |
+
"loss": -0.0046,
|
| 285 |
+
"step": 125
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.5205205205205206,
|
| 289 |
+
"grad_norm": 0.12420658767223358,
|
| 290 |
+
"grpo_mean_advantage": 2.9876827056796174e-07,
|
| 291 |
+
"grpo_mean_group_score": 0.6111599802970886,
|
| 292 |
+
"grpo_mean_kl_div": 0.0,
|
| 293 |
+
"grpo_std_advantage": 1.0496698905626545e-06,
|
| 294 |
+
"learning_rate": 4.4315423712133595e-06,
|
| 295 |
+
"loss": -0.003,
|
| 296 |
+
"step": 130
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.5405405405405406,
|
| 300 |
+
"grad_norm": 0.14342808723449707,
|
| 301 |
+
"grpo_mean_advantage": 1.5869736103013565e-07,
|
| 302 |
+
"grpo_mean_group_score": 0.5619662404060364,
|
| 303 |
+
"grpo_mean_kl_div": 0.0,
|
| 304 |
+
"grpo_std_advantage": 1.2748531617035042e-06,
|
| 305 |
+
"learning_rate": 4.378008757385222e-06,
|
| 306 |
+
"loss": 0.0154,
|
| 307 |
+
"step": 135
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.5605605605605606,
|
| 311 |
+
"grad_norm": 0.14729444682598114,
|
| 312 |
+
"grpo_mean_advantage": 3.0100346748440643e-07,
|
| 313 |
+
"grpo_mean_group_score": 0.5795454978942871,
|
| 314 |
+
"grpo_mean_kl_div": 0.0,
|
| 315 |
+
"grpo_std_advantage": 2.4499684059264837e-06,
|
| 316 |
+
"learning_rate": 4.322421568553529e-06,
|
| 317 |
+
"loss": -0.0262,
|
| 318 |
+
"step": 140
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.5805805805805806,
|
| 322 |
+
"grad_norm": 0.15249410271644592,
|
| 323 |
+
"grpo_mean_advantage": -3.233552092751779e-07,
|
| 324 |
+
"grpo_mean_group_score": 0.5804953575134277,
|
| 325 |
+
"grpo_mean_kl_div": 0.0,
|
| 326 |
+
"grpo_std_advantage": 1.248456669600273e-06,
|
| 327 |
+
"learning_rate": 4.2648415884931476e-06,
|
| 328 |
+
"loss": 0.0018,
|
| 329 |
+
"step": 145
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.6006006006006006,
|
| 333 |
+
"grad_norm": 0.1841023564338684,
|
| 334 |
+
"grpo_mean_advantage": 3.2261013416245987e-07,
|
| 335 |
+
"grpo_mean_group_score": 0.5628539323806763,
|
| 336 |
+
"grpo_mean_kl_div": 0.0,
|
| 337 |
+
"grpo_std_advantage": 1.4773489738217904e-06,
|
| 338 |
+
"learning_rate": 4.205331780066892e-06,
|
| 339 |
+
"loss": -0.017,
|
| 340 |
+
"step": 150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6206206206206206,
|
| 344 |
+
"grad_norm": 0.18597163259983063,
|
| 345 |
+
"grpo_mean_advantage": -2.5331974029541016e-07,
|
| 346 |
+
"grpo_mean_group_score": 0.5727725625038147,
|
| 347 |
+
"grpo_mean_kl_div": 0.0,
|
| 348 |
+
"grpo_std_advantage": 1.5092309695319273e-06,
|
| 349 |
+
"learning_rate": 4.1439572163765615e-06,
|
| 350 |
+
"loss": 0.0044,
|
| 351 |
+
"step": 155
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.6406406406406406,
|
| 355 |
+
"grad_norm": 0.18310388922691345,
|
| 356 |
+
"grpo_mean_advantage": -6.780028627417778e-08,
|
| 357 |
+
"grpo_mean_group_score": 0.5833909511566162,
|
| 358 |
+
"grpo_mean_kl_div": 0.0,
|
| 359 |
+
"grpo_std_advantage": 8.550978805033083e-07,
|
| 360 |
+
"learning_rate": 4.0807850096064605e-06,
|
| 361 |
+
"loss": -0.005,
|
| 362 |
+
"step": 160
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6606606606606606,
|
| 366 |
+
"grad_norm": 0.2192923128604889,
|
| 367 |
+
"grpo_mean_advantage": -5.587935447692871e-08,
|
| 368 |
+
"grpo_mean_group_score": 0.5742615461349487,
|
| 369 |
+
"grpo_mean_kl_div": 0.0,
|
| 370 |
+
"grpo_std_advantage": 3.564579174053506e-07,
|
| 371 |
+
"learning_rate": 4.015884237637206e-06,
|
| 372 |
+
"loss": -0.015,
|
| 373 |
+
"step": 165
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.6806806806806807,
|
| 377 |
+
"grad_norm": 0.16708803176879883,
|
| 378 |
+
"grpo_mean_advantage": -5.327165126800537e-07,
|
| 379 |
+
"grpo_mean_group_score": 0.5758188962936401,
|
| 380 |
+
"grpo_mean_kl_div": 0.0,
|
| 381 |
+
"grpo_std_advantage": 2.309018327650847e-06,
|
| 382 |
+
"learning_rate": 3.949325868510083e-06,
|
| 383 |
+
"loss": -0.0314,
|
| 384 |
+
"step": 170
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7007007007007007,
|
| 388 |
+
"grad_norm": 0.3401262164115906,
|
| 389 |
+
"grpo_mean_advantage": 5.863606702405377e-07,
|
| 390 |
+
"grpo_mean_group_score": 0.5767683982849121,
|
| 391 |
+
"grpo_mean_kl_div": 0.0,
|
| 392 |
+
"grpo_std_advantage": 2.4449204829579685e-06,
|
| 393 |
+
"learning_rate": 3.881182682824534e-06,
|
| 394 |
+
"loss": -0.0441,
|
| 395 |
+
"step": 175
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.7207207207207207,
|
| 399 |
+
"grad_norm": 0.1931898146867752,
|
| 400 |
+
"grpo_mean_advantage": 3.2186508747145126e-07,
|
| 401 |
+
"grpo_mean_group_score": 0.586772084236145,
|
| 402 |
+
"grpo_mean_kl_div": 0.0,
|
| 403 |
+
"grpo_std_advantage": 2.293551688126172e-06,
|
| 404 |
+
"learning_rate": 3.811529194153635e-06,
|
| 405 |
+
"loss": -0.0162,
|
| 406 |
+
"step": 180
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7407407407407407,
|
| 410 |
+
"grad_norm": 0.2537969648838043,
|
| 411 |
+
"grpo_mean_advantage": -4.470348358154297e-08,
|
| 412 |
+
"grpo_mean_group_score": 0.549396276473999,
|
| 413 |
+
"grpo_mean_kl_div": 0.0,
|
| 414 |
+
"grpo_std_advantage": 3.7067667335577426e-07,
|
| 415 |
+
"learning_rate": 3.7404415675646054e-06,
|
| 416 |
+
"loss": -0.0386,
|
| 417 |
+
"step": 185
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.7607607607607607,
|
| 421 |
+
"grad_norm": 0.20326584577560425,
|
| 422 |
+
"grpo_mean_advantage": -2.1010637851759384e-07,
|
| 423 |
+
"grpo_mean_group_score": 0.5798425078392029,
|
| 424 |
+
"grpo_mean_kl_div": 0.0,
|
| 425 |
+
"grpo_std_advantage": 1.1695076409523608e-06,
|
| 426 |
+
"learning_rate": 3.667997536333424e-06,
|
| 427 |
+
"loss": -0.037,
|
| 428 |
+
"step": 190
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.7807807807807807,
|
| 432 |
+
"grad_norm": 0.25048357248306274,
|
| 433 |
+
"grpo_mean_advantage": 1.765787658314366e-07,
|
| 434 |
+
"grpo_mean_group_score": 0.5584167838096619,
|
| 435 |
+
"grpo_mean_kl_div": 0.0,
|
| 436 |
+
"grpo_std_advantage": 2.429934738756856e-06,
|
| 437 |
+
"learning_rate": 3.59427631694463e-06,
|
| 438 |
+
"loss": -0.0292,
|
| 439 |
+
"step": 195
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.8008008008008008,
|
| 443 |
+
"grad_norm": 0.2687569260597229,
|
| 444 |
+
"grpo_mean_advantage": 1.6540289493605087e-07,
|
| 445 |
+
"grpo_mean_group_score": 0.5676193237304688,
|
| 446 |
+
"grpo_mean_kl_div": 0.0,
|
| 447 |
+
"grpo_std_advantage": 2.6342788714828203e-06,
|
| 448 |
+
"learning_rate": 3.5193585224692595e-06,
|
| 449 |
+
"loss": -0.0454,
|
| 450 |
+
"step": 200
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8208208208208209,
|
| 454 |
+
"grad_norm": 0.22301620244979858,
|
| 455 |
+
"grpo_mean_advantage": -1.0944902442133753e-06,
|
| 456 |
+
"grpo_mean_group_score": 0.5669739842414856,
|
| 457 |
+
"grpo_mean_kl_div": 0.0,
|
| 458 |
+
"grpo_std_advantage": 5.346942998585291e-06,
|
| 459 |
+
"learning_rate": 3.44332607441564e-06,
|
| 460 |
+
"loss": -0.0423,
|
| 461 |
+
"step": 205
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.8408408408408409,
|
| 465 |
+
"grad_norm": 0.3040211498737335,
|
| 466 |
+
"grpo_mean_advantage": 2.4065374759629776e-07,
|
| 467 |
+
"grpo_mean_group_score": 0.5922158360481262,
|
| 468 |
+
"grpo_mean_kl_div": 0.0,
|
| 469 |
+
"grpo_std_advantage": 1.6327536513927043e-06,
|
| 470 |
+
"learning_rate": 3.3662621131494204e-06,
|
| 471 |
+
"loss": -0.0857,
|
| 472 |
+
"step": 210
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.8608608608608609,
|
| 476 |
+
"grad_norm": 0.27231141924858093,
|
| 477 |
+
"grpo_mean_advantage": -5.21540641784668e-08,
|
| 478 |
+
"grpo_mean_group_score": 0.5473950505256653,
|
| 479 |
+
"grpo_mean_kl_div": 0.0,
|
| 480 |
+
"grpo_std_advantage": 5.847922466273303e-07,
|
| 481 |
+
"learning_rate": 3.2882509069808044e-06,
|
| 482 |
+
"loss": -0.0278,
|
| 483 |
+
"step": 215
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.8808808808808809,
|
| 487 |
+
"grad_norm": 0.3571636378765106,
|
| 488 |
+
"grpo_mean_advantage": 6.541609991472797e-07,
|
| 489 |
+
"grpo_mean_group_score": 0.5880032777786255,
|
| 490 |
+
"grpo_mean_kl_div": 0.0,
|
| 491 |
+
"grpo_std_advantage": 4.072162937518442e-06,
|
| 492 |
+
"learning_rate": 3.2093777600183873e-06,
|
| 493 |
+
"loss": -0.0727,
|
| 494 |
+
"step": 220
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9009009009009009,
|
| 498 |
+
"grad_norm": 0.306273490190506,
|
| 499 |
+
"grpo_mean_advantage": -1.2218951894737984e-07,
|
| 500 |
+
"grpo_mean_group_score": 0.5835092663764954,
|
| 501 |
+
"grpo_mean_kl_div": 0.0,
|
| 502 |
+
"grpo_std_advantage": 4.386006935419573e-07,
|
| 503 |
+
"learning_rate": 3.1297289188903705e-06,
|
| 504 |
+
"loss": -0.0464,
|
| 505 |
+
"step": 225
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.9209209209209209,
|
| 509 |
+
"grad_norm": 0.2700377106666565,
|
| 510 |
+
"grpo_mean_advantage": 1.7605722177904681e-06,
|
| 511 |
+
"grpo_mean_group_score": 0.5394966006278992,
|
| 512 |
+
"grpo_mean_kl_div": 0.0,
|
| 513 |
+
"grpo_std_advantage": 8.007580618141219e-06,
|
| 514 |
+
"learning_rate": 3.049391478435133e-06,
|
| 515 |
+
"loss": -0.0295,
|
| 516 |
+
"step": 230
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9409409409409409,
|
| 520 |
+
"grad_norm": 0.39531761407852173,
|
| 521 |
+
"grpo_mean_advantage": -3.3080578987210174e-07,
|
| 522 |
+
"grpo_mean_group_score": 0.5687432289123535,
|
| 523 |
+
"grpo_mean_kl_div": 0.0,
|
| 524 |
+
"grpo_std_advantage": 1.551636614749441e-06,
|
| 525 |
+
"learning_rate": 2.9684532864643123e-06,
|
| 526 |
+
"loss": -0.031,
|
| 527 |
+
"step": 235
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.960960960960961,
|
| 531 |
+
"grad_norm": 0.5987040996551514,
|
| 532 |
+
"grpo_mean_advantage": 2.712011450967111e-07,
|
| 533 |
+
"grpo_mean_group_score": 0.5550583600997925,
|
| 534 |
+
"grpo_mean_kl_div": 0.0,
|
| 535 |
+
"grpo_std_advantage": 1.4400844747797237e-06,
|
| 536 |
+
"learning_rate": 2.887002847702504e-06,
|
| 537 |
+
"loss": -0.0789,
|
| 538 |
+
"step": 240
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.980980980980981,
|
| 542 |
+
"grad_norm": 0.5680716037750244,
|
| 543 |
+
"grpo_mean_advantage": -3.2857059295565705e-07,
|
| 544 |
+
"grpo_mean_group_score": 0.558111310005188,
|
| 545 |
+
"grpo_mean_kl_div": 0.0,
|
| 546 |
+
"grpo_std_advantage": 2.105091425619321e-06,
|
| 547 |
+
"learning_rate": 2.8051292270086506e-06,
|
| 548 |
+
"loss": -0.1131,
|
| 549 |
+
"step": 245
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.0,
|
| 553 |
+
"grad_norm": 0.6204046010971069,
|
| 554 |
+
"grpo_mean_advantage": 4.470348358154297e-08,
|
| 555 |
+
"grpo_mean_group_score": 0.6196198463439941,
|
| 556 |
+
"grpo_mean_kl_div": 0.0,
|
| 557 |
+
"grpo_std_advantage": 5.315724820320611e-07,
|
| 558 |
+
"learning_rate": 2.722921951984927e-06,
|
| 559 |
+
"loss": -0.2232,
|
| 560 |
+
"step": 250
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.02002002002002,
|
| 564 |
+
"grad_norm": 0.8389026522636414,
|
| 565 |
+
"grpo_mean_advantage": 9.290873776990338e-07,
|
| 566 |
+
"grpo_mean_group_score": 0.582168459892273,
|
| 567 |
+
"grpo_mean_kl_div": 0.0,
|
| 568 |
+
"grpo_std_advantage": 4.219644324621186e-06,
|
| 569 |
+
"learning_rate": 2.640470915079614e-06,
|
| 570 |
+
"loss": -0.1363,
|
| 571 |
+
"step": 255
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.04004004004004,
|
| 575 |
+
"grad_norm": 0.9067686796188354,
|
| 576 |
+
"grpo_mean_advantage": 2.533197474008375e-08,
|
| 577 |
+
"grpo_mean_group_score": 0.5551307797431946,
|
| 578 |
+
"grpo_mean_kl_div": 0.0,
|
| 579 |
+
"grpo_std_advantage": 1.6600588992332632e-07,
|
| 580 |
+
"learning_rate": 2.557866275291035e-06,
|
| 581 |
+
"loss": -0.1868,
|
| 582 |
+
"step": 260
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.06006006006006,
|
| 586 |
+
"grad_norm": 0.9277902841567993,
|
| 587 |
+
"grpo_mean_advantage": -5.662441182607836e-08,
|
| 588 |
+
"grpo_mean_group_score": 0.535040020942688,
|
| 589 |
+
"grpo_mean_kl_div": 0.0,
|
| 590 |
+
"grpo_std_advantage": 1.0909400316450046e-06,
|
| 591 |
+
"learning_rate": 2.4751983595800093e-06,
|
| 592 |
+
"loss": -0.1792,
|
| 593 |
+
"step": 265
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.08008008008008,
|
| 597 |
+
"grad_norm": 1.0715463161468506,
|
| 598 |
+
"grpo_mean_advantage": -9.536743306171047e-08,
|
| 599 |
+
"grpo_mean_group_score": 0.5673571825027466,
|
| 600 |
+
"grpo_mean_kl_div": 0.0,
|
| 601 |
+
"grpo_std_advantage": 5.838213610331877e-07,
|
| 602 |
+
"learning_rate": 2.392557564098649e-06,
|
| 603 |
+
"loss": -0.1691,
|
| 604 |
+
"step": 270
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.1001001001001,
|
| 608 |
+
"grad_norm": 0.7759184837341309,
|
| 609 |
+
"grpo_mean_advantage": 3.278255533700758e-08,
|
| 610 |
+
"grpo_mean_group_score": 0.5874732732772827,
|
| 611 |
+
"grpo_mean_kl_div": 0.0,
|
| 612 |
+
"grpo_std_advantage": 9.317170679423725e-07,
|
| 613 |
+
"learning_rate": 2.3100342553434924e-06,
|
| 614 |
+
"loss": -0.1655,
|
| 615 |
+
"step": 275
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.12012012012012,
|
| 619 |
+
"grad_norm": 0.9387398958206177,
|
| 620 |
+
"grpo_mean_advantage": -1.206994113545079e-07,
|
| 621 |
+
"grpo_mean_group_score": 0.5569106340408325,
|
| 622 |
+
"grpo_mean_kl_div": 0.0,
|
| 623 |
+
"grpo_std_advantage": 6.201085511747806e-07,
|
| 624 |
+
"learning_rate": 2.2277186713410688e-06,
|
| 625 |
+
"loss": -0.1821,
|
| 626 |
+
"step": 280
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.14014014014014,
|
| 630 |
+
"grad_norm": 1.6132302284240723,
|
| 631 |
+
"grpo_mean_advantage": 4.470348358154297e-08,
|
| 632 |
+
"grpo_mean_group_score": 0.5578873157501221,
|
| 633 |
+
"grpo_mean_kl_div": 0.0,
|
| 634 |
+
"grpo_std_advantage": 6.115651558502577e-07,
|
| 635 |
+
"learning_rate": 2.1457008229739395e-06,
|
| 636 |
+
"loss": -0.2102,
|
| 637 |
+
"step": 285
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.16016016016016,
|
| 641 |
+
"grad_norm": 0.8679026961326599,
|
| 642 |
+
"grpo_mean_advantage": -3.3453108017056365e-07,
|
| 643 |
+
"grpo_mean_group_score": 0.5735999345779419,
|
| 644 |
+
"grpo_mean_kl_div": 0.0,
|
| 645 |
+
"grpo_std_advantage": 3.5326345368957845e-06,
|
| 646 |
+
"learning_rate": 2.0640703955551214e-06,
|
| 647 |
+
"loss": -0.2937,
|
| 648 |
+
"step": 290
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.1801801801801801,
|
| 652 |
+
"grad_norm": 1.0550166368484497,
|
| 653 |
+
"grpo_mean_advantage": -1.110136480519941e-07,
|
| 654 |
+
"grpo_mean_group_score": 0.5626259446144104,
|
| 655 |
+
"grpo_mean_kl_div": 0.0,
|
| 656 |
+
"grpo_std_advantage": 4.731904823529476e-07,
|
| 657 |
+
"learning_rate": 1.9829166507585084e-06,
|
| 658 |
+
"loss": -0.2598,
|
| 659 |
+
"step": 295
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 1.2002002002002001,
|
| 663 |
+
"grad_norm": 1.2819372415542603,
|
| 664 |
+
"grpo_mean_advantage": -5.08874677507265e-07,
|
| 665 |
+
"grpo_mean_group_score": 0.5463050603866577,
|
| 666 |
+
"grpo_mean_kl_div": 0.0,
|
| 667 |
+
"grpo_std_advantage": 1.840126174101897e-06,
|
| 668 |
+
"learning_rate": 1.90232832901255e-06,
|
| 669 |
+
"loss": -0.2546,
|
| 670 |
+
"step": 300
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.2202202202202201,
|
| 674 |
+
"grad_norm": 1.0188143253326416,
|
| 675 |
+
"grpo_mean_advantage": 1.01327898960335e-07,
|
| 676 |
+
"grpo_mean_group_score": 0.5352144241333008,
|
| 677 |
+
"grpo_mean_kl_div": 0.0,
|
| 678 |
+
"grpo_std_advantage": 7.798533943059738e-07,
|
| 679 |
+
"learning_rate": 1.82239355246389e-06,
|
| 680 |
+
"loss": -0.1809,
|
| 681 |
+
"step": 305
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 1.2402402402402402,
|
| 685 |
+
"grad_norm": 2.0709052085876465,
|
| 686 |
+
"grpo_mean_advantage": 1.341104507446289e-07,
|
| 687 |
+
"grpo_mean_group_score": 0.5547868013381958,
|
| 688 |
+
"grpo_mean_kl_div": 0.0,
|
| 689 |
+
"grpo_std_advantage": 7.821902840987605e-07,
|
| 690 |
+
"learning_rate": 1.7431997286170923e-06,
|
| 691 |
+
"loss": -0.3559,
|
| 692 |
+
"step": 310
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.2602602602602602,
|
| 696 |
+
"grad_norm": 1.8516215085983276,
|
| 697 |
+
"grpo_mean_advantage": 9.015202806494926e-08,
|
| 698 |
+
"grpo_mean_group_score": 0.5859472751617432,
|
| 699 |
+
"grpo_mean_kl_div": 0.0,
|
| 700 |
+
"grpo_std_advantage": 1.0693488547985908e-06,
|
| 701 |
+
"learning_rate": 1.6648334547558227e-06,
|
| 702 |
+
"loss": -0.3874,
|
| 703 |
+
"step": 315
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.2802802802802802,
|
| 707 |
+
"grad_norm": 1.283104419708252,
|
| 708 |
+
"grpo_mean_advantage": -2.443790378947597e-07,
|
| 709 |
+
"grpo_mean_group_score": 0.5751550793647766,
|
| 710 |
+
"grpo_mean_kl_div": 0.0,
|
| 711 |
+
"grpo_std_advantage": 1.183122208203713e-06,
|
| 712 |
+
"learning_rate": 1.5873804232499862e-06,
|
| 713 |
+
"loss": -0.3467,
|
| 714 |
+
"step": 320
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3003003003003002,
|
| 718 |
+
"grad_norm": 1.4108576774597168,
|
| 719 |
+
"grpo_mean_advantage": -6.705522537231445e-08,
|
| 720 |
+
"grpo_mean_group_score": 0.5497723817825317,
|
| 721 |
+
"grpo_mean_kl_div": 0.0,
|
| 722 |
+
"grpo_std_advantage": 6.109748937888071e-07,
|
| 723 |
+
"learning_rate": 1.51092532785238e-06,
|
| 724 |
+
"loss": -0.1703,
|
| 725 |
+
"step": 325
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.3203203203203202,
|
| 729 |
+
"grad_norm": 1.0421361923217773,
|
| 730 |
+
"grpo_mean_advantage": -1.639127766850379e-08,
|
| 731 |
+
"grpo_mean_group_score": 0.55989670753479,
|
| 732 |
+
"grpo_mean_kl_div": 0.0,
|
| 733 |
+
"grpo_std_advantage": 5.529495297196263e-07,
|
| 734 |
+
"learning_rate": 1.4355517710873184e-06,
|
| 735 |
+
"loss": -0.2918,
|
| 736 |
+
"step": 330
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.3403403403403402,
|
| 740 |
+
"grad_norm": 1.3465828895568848,
|
| 741 |
+
"grpo_mean_advantage": 4.418194237132411e-07,
|
| 742 |
+
"grpo_mean_group_score": 0.5809233784675598,
|
| 743 |
+
"grpo_mean_kl_div": 0.0,
|
| 744 |
+
"grpo_std_advantage": 2.9275292945385445e-06,
|
| 745 |
+
"learning_rate": 1.361342172832502e-06,
|
| 746 |
+
"loss": -0.3069,
|
| 747 |
+
"step": 335
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.3603603603603602,
|
| 751 |
+
"grad_norm": 1.1959459781646729,
|
| 752 |
+
"grpo_mean_advantage": 9.685754776000977e-08,
|
| 753 |
+
"grpo_mean_group_score": 0.5568087100982666,
|
| 754 |
+
"grpo_mean_kl_div": 0.0,
|
| 755 |
+
"grpo_std_advantage": 3.754235251562932e-07,
|
| 756 |
+
"learning_rate": 1.2883776801940884e-06,
|
| 757 |
+
"loss": -0.5594,
|
| 758 |
+
"step": 340
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.3803803803803802,
|
| 762 |
+
"grad_norm": 1.8967422246932983,
|
| 763 |
+
"grpo_mean_advantage": -2.384185791015625e-07,
|
| 764 |
+
"grpo_mean_group_score": 0.5655568838119507,
|
| 765 |
+
"grpo_mean_kl_div": 0.0,
|
| 766 |
+
"grpo_std_advantage": 6.821086913078034e-07,
|
| 767 |
+
"learning_rate": 1.216738078773522e-06,
|
| 768 |
+
"loss": -0.4102,
|
| 769 |
+
"step": 345
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.4004004004004005,
|
| 773 |
+
"grad_norm": 2.221132755279541,
|
| 774 |
+
"grpo_mean_advantage": -8.717179156292332e-08,
|
| 775 |
+
"grpo_mean_group_score": 0.6089578866958618,
|
| 776 |
+
"grpo_mean_kl_div": 0.0,
|
| 777 |
+
"grpo_std_advantage": 2.500940354366321e-06,
|
| 778 |
+
"learning_rate": 1.146501705423155e-06,
|
| 779 |
+
"loss": -0.338,
|
| 780 |
+
"step": 350
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.4204204204204205,
|
| 784 |
+
"grad_norm": 2.3640377521514893,
|
| 785 |
+
"grpo_mean_advantage": 2.1606683731079102e-07,
|
| 786 |
+
"grpo_mean_group_score": 0.6129671335220337,
|
| 787 |
+
"grpo_mean_kl_div": 0.0,
|
| 788 |
+
"grpo_std_advantage": 1.4568390724889468e-06,
|
| 789 |
+
"learning_rate": 1.0777453625860474e-06,
|
| 790 |
+
"loss": -0.4985,
|
| 791 |
+
"step": 355
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.4404404404404405,
|
| 795 |
+
"grad_norm": 1.9084734916687012,
|
| 796 |
+
"grpo_mean_advantage": -3.725290298461914e-09,
|
| 797 |
+
"grpo_mean_group_score": 0.5562310814857483,
|
| 798 |
+
"grpo_mean_kl_div": 0.0,
|
| 799 |
+
"grpo_std_advantage": 2.965894054796081e-06,
|
| 800 |
+
"learning_rate": 1.0105442343136184e-06,
|
| 801 |
+
"loss": -0.4347,
|
| 802 |
+
"step": 360
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4604604604604605,
|
| 806 |
+
"grad_norm": 1.6063904762268066,
|
| 807 |
+
"grpo_mean_advantage": 4.313886279305734e-07,
|
| 808 |
+
"grpo_mean_group_score": 0.5884170532226562,
|
| 809 |
+
"grpo_mean_kl_div": 0.0,
|
| 810 |
+
"grpo_std_advantage": 1.9621948013082147e-06,
|
| 811 |
+
"learning_rate": 9.449718040529987e-07,
|
| 812 |
+
"loss": -0.6217,
|
| 813 |
+
"step": 365
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.4804804804804805,
|
| 817 |
+
"grad_norm": 2.114664077758789,
|
| 818 |
+
"grpo_mean_advantage": 2.0489096641540527e-07,
|
| 819 |
+
"grpo_mean_group_score": 0.5795440673828125,
|
| 820 |
+
"grpo_mean_kl_div": 0.0,
|
| 821 |
+
"grpo_std_advantage": 1.0235522722723545e-06,
|
| 822 |
+
"learning_rate": 8.810997742939531e-07,
|
| 823 |
+
"loss": -0.5364,
|
| 824 |
+
"step": 370
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.5005005005005005,
|
| 828 |
+
"grad_norm": 1.8450465202331543,
|
| 829 |
+
"grpo_mean_advantage": -1.4185905001795618e-06,
|
| 830 |
+
"grpo_mean_group_score": 0.5607603788375854,
|
| 831 |
+
"grpo_mean_kl_div": 0.0,
|
| 832 |
+
"grpo_std_advantage": 1.0947338523692451e-05,
|
| 833 |
+
"learning_rate": 8.189979881632634e-07,
|
| 834 |
+
"loss": -0.4798,
|
| 835 |
+
"step": 375
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.5205205205205206,
|
| 839 |
+
"grad_norm": 2.673438787460327,
|
| 840 |
+
"grpo_mean_advantage": -1.758337049295733e-07,
|
| 841 |
+
"grpo_mean_group_score": 0.5381432771682739,
|
| 842 |
+
"grpo_mean_kl_div": 0.0,
|
| 843 |
+
"grpo_std_advantage": 9.663675655247062e-07,
|
| 844 |
+
"learning_rate": 7.587343530522945e-07,
|
| 845 |
+
"loss": -0.4805,
|
| 846 |
+
"step": 380
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.5405405405405406,
|
| 850 |
+
"grad_norm": 2.2263550758361816,
|
| 851 |
+
"grpo_mean_advantage": -6.973743325033865e-07,
|
| 852 |
+
"grpo_mean_group_score": 0.5528443455696106,
|
| 853 |
+
"grpo_mean_kl_div": 0.0,
|
| 854 |
+
"grpo_std_advantage": 4.341973180999048e-06,
|
| 855 |
+
"learning_rate": 7.003747663612581e-07,
|
| 856 |
+
"loss": -0.433,
|
| 857 |
+
"step": 385
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 1.5605605605605606,
|
| 861 |
+
"grad_norm": 2.3657093048095703,
|
| 862 |
+
"grpo_mean_advantage": 1.7881394143159923e-08,
|
| 863 |
+
"grpo_mean_group_score": 0.6091476678848267,
|
| 864 |
+
"grpo_mean_kl_div": 0.0,
|
| 865 |
+
"grpo_std_advantage": 1.3004198251564958e-07,
|
| 866 |
+
"learning_rate": 6.439830434413754e-07,
|
| 867 |
+
"loss": -0.6021,
|
| 868 |
+
"step": 390
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.5805805805805806,
|
| 872 |
+
"grad_norm": 1.9847129583358765,
|
| 873 |
+
"grpo_mean_advantage": 3.4868716625169327e-07,
|
| 874 |
+
"grpo_mean_group_score": 0.5397372245788574,
|
| 875 |
+
"grpo_mean_kl_div": 0.0,
|
| 876 |
+
"grpo_std_advantage": 2.059372718576924e-06,
|
| 877 |
+
"learning_rate": 5.896208478137222e-07,
|
| 878 |
+
"loss": -0.5595,
|
| 879 |
+
"step": 395
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.6006006006006006,
|
| 883 |
+
"grad_norm": 2.922114133834839,
|
| 884 |
+
"grpo_mean_advantage": -2.1636485598719446e-06,
|
| 885 |
+
"grpo_mean_group_score": 0.5873125195503235,
|
| 886 |
+
"grpo_mean_kl_div": 0.0,
|
| 887 |
+
"grpo_std_advantage": 9.725940799398813e-06,
|
| 888 |
+
"learning_rate": 5.373476237410808e-07,
|
| 889 |
+
"loss": -0.5592,
|
| 890 |
+
"step": 400
|
| 891 |
+
}
|
| 892 |
+
],
|
| 893 |
+
"logging_steps": 5,
|
| 894 |
+
"max_steps": 500,
|
| 895 |
+
"num_input_tokens_seen": 0,
|
| 896 |
+
"num_train_epochs": 2,
|
| 897 |
+
"save_steps": 100,
|
| 898 |
+
"stateful_callbacks": {
|
| 899 |
+
"TrainerControl": {
|
| 900 |
+
"args": {
|
| 901 |
+
"should_epoch_stop": false,
|
| 902 |
+
"should_evaluate": false,
|
| 903 |
+
"should_log": false,
|
| 904 |
+
"should_save": true,
|
| 905 |
+
"should_training_stop": false
|
| 906 |
+
},
|
| 907 |
+
"attributes": {}
|
| 908 |
+
}
|
| 909 |
+
},
|
| 910 |
+
"total_flos": 0.0,
|
| 911 |
+
"train_batch_size": 1,
|
| 912 |
+
"trial_name": null,
|
| 913 |
+
"trial_params": null
|
| 914 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a4a4e48ed61b7c96f3bd2836ac828013a311834ab8a9542ea461fe1ff953396b
|
| 3 |
+
size 5496
|
grpo_qwen_14B_v2/checkpoints/checkpoint-400/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/README.md
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 3 |
+
library_name: peft
|
| 4 |
+
pipeline_tag: text-generation
|
| 5 |
+
tags:
|
| 6 |
+
- base_model:adapter:/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 7 |
+
- lora
|
| 8 |
+
- transformers
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Model ID
|
| 12 |
+
|
| 13 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
## Model Details
|
| 18 |
+
|
| 19 |
+
### Model Description
|
| 20 |
+
|
| 21 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
- **Developed by:** [More Information Needed]
|
| 26 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 27 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 28 |
+
- **Model type:** [More Information Needed]
|
| 29 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 30 |
+
- **License:** [More Information Needed]
|
| 31 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 32 |
+
|
| 33 |
+
### Model Sources [optional]
|
| 34 |
+
|
| 35 |
+
<!-- Provide the basic links for the model. -->
|
| 36 |
+
|
| 37 |
+
- **Repository:** [More Information Needed]
|
| 38 |
+
- **Paper [optional]:** [More Information Needed]
|
| 39 |
+
- **Demo [optional]:** [More Information Needed]
|
| 40 |
+
|
| 41 |
+
## Uses
|
| 42 |
+
|
| 43 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 44 |
+
|
| 45 |
+
### Direct Use
|
| 46 |
+
|
| 47 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 48 |
+
|
| 49 |
+
[More Information Needed]
|
| 50 |
+
|
| 51 |
+
### Downstream Use [optional]
|
| 52 |
+
|
| 53 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 54 |
+
|
| 55 |
+
[More Information Needed]
|
| 56 |
+
|
| 57 |
+
### Out-of-Scope Use
|
| 58 |
+
|
| 59 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 60 |
+
|
| 61 |
+
[More Information Needed]
|
| 62 |
+
|
| 63 |
+
## Bias, Risks, and Limitations
|
| 64 |
+
|
| 65 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 66 |
+
|
| 67 |
+
[More Information Needed]
|
| 68 |
+
|
| 69 |
+
### Recommendations
|
| 70 |
+
|
| 71 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 72 |
+
|
| 73 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 74 |
+
|
| 75 |
+
## How to Get Started with the Model
|
| 76 |
+
|
| 77 |
+
Use the code below to get started with the model.
|
| 78 |
+
|
| 79 |
+
[More Information Needed]
|
| 80 |
+
|
| 81 |
+
## Training Details
|
| 82 |
+
|
| 83 |
+
### Training Data
|
| 84 |
+
|
| 85 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 86 |
+
|
| 87 |
+
[More Information Needed]
|
| 88 |
+
|
| 89 |
+
### Training Procedure
|
| 90 |
+
|
| 91 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 92 |
+
|
| 93 |
+
#### Preprocessing [optional]
|
| 94 |
+
|
| 95 |
+
[More Information Needed]
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
#### Training Hyperparameters
|
| 99 |
+
|
| 100 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 101 |
+
|
| 102 |
+
#### Speeds, Sizes, Times [optional]
|
| 103 |
+
|
| 104 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 105 |
+
|
| 106 |
+
[More Information Needed]
|
| 107 |
+
|
| 108 |
+
## Evaluation
|
| 109 |
+
|
| 110 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 111 |
+
|
| 112 |
+
### Testing Data, Factors & Metrics
|
| 113 |
+
|
| 114 |
+
#### Testing Data
|
| 115 |
+
|
| 116 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 117 |
+
|
| 118 |
+
[More Information Needed]
|
| 119 |
+
|
| 120 |
+
#### Factors
|
| 121 |
+
|
| 122 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 123 |
+
|
| 124 |
+
[More Information Needed]
|
| 125 |
+
|
| 126 |
+
#### Metrics
|
| 127 |
+
|
| 128 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 129 |
+
|
| 130 |
+
[More Information Needed]
|
| 131 |
+
|
| 132 |
+
### Results
|
| 133 |
+
|
| 134 |
+
[More Information Needed]
|
| 135 |
+
|
| 136 |
+
#### Summary
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
|
| 140 |
+
## Model Examination [optional]
|
| 141 |
+
|
| 142 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 143 |
+
|
| 144 |
+
[More Information Needed]
|
| 145 |
+
|
| 146 |
+
## Environmental Impact
|
| 147 |
+
|
| 148 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 149 |
+
|
| 150 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 151 |
+
|
| 152 |
+
- **Hardware Type:** [More Information Needed]
|
| 153 |
+
- **Hours used:** [More Information Needed]
|
| 154 |
+
- **Cloud Provider:** [More Information Needed]
|
| 155 |
+
- **Compute Region:** [More Information Needed]
|
| 156 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 157 |
+
|
| 158 |
+
## Technical Specifications [optional]
|
| 159 |
+
|
| 160 |
+
### Model Architecture and Objective
|
| 161 |
+
|
| 162 |
+
[More Information Needed]
|
| 163 |
+
|
| 164 |
+
### Compute Infrastructure
|
| 165 |
+
|
| 166 |
+
[More Information Needed]
|
| 167 |
+
|
| 168 |
+
#### Hardware
|
| 169 |
+
|
| 170 |
+
[More Information Needed]
|
| 171 |
+
|
| 172 |
+
#### Software
|
| 173 |
+
|
| 174 |
+
[More Information Needed]
|
| 175 |
+
|
| 176 |
+
## Citation [optional]
|
| 177 |
+
|
| 178 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 179 |
+
|
| 180 |
+
**BibTeX:**
|
| 181 |
+
|
| 182 |
+
[More Information Needed]
|
| 183 |
+
|
| 184 |
+
**APA:**
|
| 185 |
+
|
| 186 |
+
[More Information Needed]
|
| 187 |
+
|
| 188 |
+
## Glossary [optional]
|
| 189 |
+
|
| 190 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 191 |
+
|
| 192 |
+
[More Information Needed]
|
| 193 |
+
|
| 194 |
+
## More Information [optional]
|
| 195 |
+
|
| 196 |
+
[More Information Needed]
|
| 197 |
+
|
| 198 |
+
## Model Card Authors [optional]
|
| 199 |
+
|
| 200 |
+
[More Information Needed]
|
| 201 |
+
|
| 202 |
+
## Model Card Contact
|
| 203 |
+
|
| 204 |
+
[More Information Needed]
|
| 205 |
+
### Framework versions
|
| 206 |
+
|
| 207 |
+
- PEFT 0.18.0
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/adapter_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": null,
|
| 6 |
+
"base_model_name_or_path": "/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2",
|
| 7 |
+
"bias": "none",
|
| 8 |
+
"corda_config": null,
|
| 9 |
+
"ensure_weight_tying": false,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 32,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"peft_version": "0.18.0",
|
| 27 |
+
"qalora_group_size": 16,
|
| 28 |
+
"r": 16,
|
| 29 |
+
"rank_pattern": {},
|
| 30 |
+
"revision": null,
|
| 31 |
+
"target_modules": [
|
| 32 |
+
"q_proj",
|
| 33 |
+
"o_proj",
|
| 34 |
+
"v_proj",
|
| 35 |
+
"up_proj",
|
| 36 |
+
"gate_proj",
|
| 37 |
+
"down_proj",
|
| 38 |
+
"k_proj"
|
| 39 |
+
],
|
| 40 |
+
"target_parameters": null,
|
| 41 |
+
"task_type": "CAUSAL_LM",
|
| 42 |
+
"trainable_token_indices": null,
|
| 43 |
+
"use_dora": false,
|
| 44 |
+
"use_qalora": false,
|
| 45 |
+
"use_rslora": false
|
| 46 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cfddfaff53a19bede615799a30a42fe443ed6813a68f14c0a605ed1e0b75aa7a
|
| 3 |
+
size 275341720
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/added_tokens.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"</tool_call>": 151658,
|
| 3 |
+
"<tool_call>": 151657,
|
| 4 |
+
"<|box_end|>": 151649,
|
| 5 |
+
"<|box_start|>": 151648,
|
| 6 |
+
"<|endoftext|>": 151643,
|
| 7 |
+
"<|file_sep|>": 151664,
|
| 8 |
+
"<|fim_middle|>": 151660,
|
| 9 |
+
"<|fim_pad|>": 151662,
|
| 10 |
+
"<|fim_prefix|>": 151659,
|
| 11 |
+
"<|fim_suffix|>": 151661,
|
| 12 |
+
"<|im_end|>": 151645,
|
| 13 |
+
"<|im_start|>": 151644,
|
| 14 |
+
"<|image_pad|>": 151655,
|
| 15 |
+
"<|object_ref_end|>": 151647,
|
| 16 |
+
"<|object_ref_start|>": 151646,
|
| 17 |
+
"<|quad_end|>": 151651,
|
| 18 |
+
"<|quad_start|>": 151650,
|
| 19 |
+
"<|repo_name|>": 151663,
|
| 20 |
+
"<|video_pad|>": 151656,
|
| 21 |
+
"<|vision_end|>": 151653,
|
| 22 |
+
"<|vision_pad|>": 151654,
|
| 23 |
+
"<|vision_start|>": 151652
|
| 24 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/chat_template.jinja
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- if tools %}
|
| 2 |
+
{{- '<|im_start|>system\n' }}
|
| 3 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 4 |
+
{{- messages[0]['content'] }}
|
| 5 |
+
{%- else %}
|
| 6 |
+
{{- 'You are a helpful assistant.' }}
|
| 7 |
+
{%- endif %}
|
| 8 |
+
{{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
|
| 9 |
+
{%- for tool in tools %}
|
| 10 |
+
{{- "\n" }}
|
| 11 |
+
{{- tool | tojson }}
|
| 12 |
+
{%- endfor %}
|
| 13 |
+
{{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
|
| 14 |
+
{%- else %}
|
| 15 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 16 |
+
{{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
|
| 17 |
+
{%- else %}
|
| 18 |
+
{{- '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}
|
| 19 |
+
{%- endif %}
|
| 20 |
+
{%- endif %}
|
| 21 |
+
{%- for message in messages %}
|
| 22 |
+
{%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
|
| 23 |
+
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
|
| 24 |
+
{%- elif message.role == "assistant" %}
|
| 25 |
+
{{- '<|im_start|>' + message.role }}
|
| 26 |
+
{%- if message.content %}
|
| 27 |
+
{{- '\n' + message.content }}
|
| 28 |
+
{%- endif %}
|
| 29 |
+
{%- for tool_call in message.tool_calls %}
|
| 30 |
+
{%- if tool_call.function is defined %}
|
| 31 |
+
{%- set tool_call = tool_call.function %}
|
| 32 |
+
{%- endif %}
|
| 33 |
+
{{- '\n<tool_call>\n{"name": "' }}
|
| 34 |
+
{{- tool_call.name }}
|
| 35 |
+
{{- '", "arguments": ' }}
|
| 36 |
+
{{- tool_call.arguments | tojson }}
|
| 37 |
+
{{- '}\n</tool_call>' }}
|
| 38 |
+
{%- endfor %}
|
| 39 |
+
{{- '<|im_end|>\n' }}
|
| 40 |
+
{%- elif message.role == "tool" %}
|
| 41 |
+
{%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
|
| 42 |
+
{{- '<|im_start|>user' }}
|
| 43 |
+
{%- endif %}
|
| 44 |
+
{{- '\n<tool_response>\n' }}
|
| 45 |
+
{{- message.content }}
|
| 46 |
+
{{- '\n</tool_response>' }}
|
| 47 |
+
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
|
| 48 |
+
{{- '<|im_end|>\n' }}
|
| 49 |
+
{%- endif %}
|
| 50 |
+
{%- endif %}
|
| 51 |
+
{%- endfor %}
|
| 52 |
+
{%- if add_generation_prompt %}
|
| 53 |
+
{{- '<|im_start|>assistant\n' }}
|
| 54 |
+
{%- endif %}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d70395226e3da425d344a5d58bb1d49f495aa05919324a7452c67daccffb659e
|
| 3 |
+
size 551070514
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:140a5b8faae8f6aa3d7c8abd94812eb021a794706417ce578c4228067f08f646
|
| 3 |
+
size 14244
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:622e95a06e3b8ce556346f153f9b954936262c4f61ed83e89373f1300da67c29
|
| 3 |
+
size 1064
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/special_tokens_map.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>",
|
| 5 |
+
"<|object_ref_start|>",
|
| 6 |
+
"<|object_ref_end|>",
|
| 7 |
+
"<|box_start|>",
|
| 8 |
+
"<|box_end|>",
|
| 9 |
+
"<|quad_start|>",
|
| 10 |
+
"<|quad_end|>",
|
| 11 |
+
"<|vision_start|>",
|
| 12 |
+
"<|vision_end|>",
|
| 13 |
+
"<|vision_pad|>",
|
| 14 |
+
"<|image_pad|>",
|
| 15 |
+
"<|video_pad|>"
|
| 16 |
+
],
|
| 17 |
+
"eos_token": {
|
| 18 |
+
"content": "<|endoftext|>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
},
|
| 24 |
+
"pad_token": {
|
| 25 |
+
"content": "<|endoftext|>",
|
| 26 |
+
"lstrip": false,
|
| 27 |
+
"normalized": false,
|
| 28 |
+
"rstrip": false,
|
| 29 |
+
"single_word": false
|
| 30 |
+
}
|
| 31 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
|
| 3 |
+
size 11421896
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/tokenizer_config.json
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"151643": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"151644": {
|
| 14 |
+
"content": "<|im_start|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"151645": {
|
| 22 |
+
"content": "<|im_end|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"151646": {
|
| 30 |
+
"content": "<|object_ref_start|>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"151647": {
|
| 38 |
+
"content": "<|object_ref_end|>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"151648": {
|
| 46 |
+
"content": "<|box_start|>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"151649": {
|
| 54 |
+
"content": "<|box_end|>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
},
|
| 61 |
+
"151650": {
|
| 62 |
+
"content": "<|quad_start|>",
|
| 63 |
+
"lstrip": false,
|
| 64 |
+
"normalized": false,
|
| 65 |
+
"rstrip": false,
|
| 66 |
+
"single_word": false,
|
| 67 |
+
"special": true
|
| 68 |
+
},
|
| 69 |
+
"151651": {
|
| 70 |
+
"content": "<|quad_end|>",
|
| 71 |
+
"lstrip": false,
|
| 72 |
+
"normalized": false,
|
| 73 |
+
"rstrip": false,
|
| 74 |
+
"single_word": false,
|
| 75 |
+
"special": true
|
| 76 |
+
},
|
| 77 |
+
"151652": {
|
| 78 |
+
"content": "<|vision_start|>",
|
| 79 |
+
"lstrip": false,
|
| 80 |
+
"normalized": false,
|
| 81 |
+
"rstrip": false,
|
| 82 |
+
"single_word": false,
|
| 83 |
+
"special": true
|
| 84 |
+
},
|
| 85 |
+
"151653": {
|
| 86 |
+
"content": "<|vision_end|>",
|
| 87 |
+
"lstrip": false,
|
| 88 |
+
"normalized": false,
|
| 89 |
+
"rstrip": false,
|
| 90 |
+
"single_word": false,
|
| 91 |
+
"special": true
|
| 92 |
+
},
|
| 93 |
+
"151654": {
|
| 94 |
+
"content": "<|vision_pad|>",
|
| 95 |
+
"lstrip": false,
|
| 96 |
+
"normalized": false,
|
| 97 |
+
"rstrip": false,
|
| 98 |
+
"single_word": false,
|
| 99 |
+
"special": true
|
| 100 |
+
},
|
| 101 |
+
"151655": {
|
| 102 |
+
"content": "<|image_pad|>",
|
| 103 |
+
"lstrip": false,
|
| 104 |
+
"normalized": false,
|
| 105 |
+
"rstrip": false,
|
| 106 |
+
"single_word": false,
|
| 107 |
+
"special": true
|
| 108 |
+
},
|
| 109 |
+
"151656": {
|
| 110 |
+
"content": "<|video_pad|>",
|
| 111 |
+
"lstrip": false,
|
| 112 |
+
"normalized": false,
|
| 113 |
+
"rstrip": false,
|
| 114 |
+
"single_word": false,
|
| 115 |
+
"special": true
|
| 116 |
+
},
|
| 117 |
+
"151657": {
|
| 118 |
+
"content": "<tool_call>",
|
| 119 |
+
"lstrip": false,
|
| 120 |
+
"normalized": false,
|
| 121 |
+
"rstrip": false,
|
| 122 |
+
"single_word": false,
|
| 123 |
+
"special": false
|
| 124 |
+
},
|
| 125 |
+
"151658": {
|
| 126 |
+
"content": "</tool_call>",
|
| 127 |
+
"lstrip": false,
|
| 128 |
+
"normalized": false,
|
| 129 |
+
"rstrip": false,
|
| 130 |
+
"single_word": false,
|
| 131 |
+
"special": false
|
| 132 |
+
},
|
| 133 |
+
"151659": {
|
| 134 |
+
"content": "<|fim_prefix|>",
|
| 135 |
+
"lstrip": false,
|
| 136 |
+
"normalized": false,
|
| 137 |
+
"rstrip": false,
|
| 138 |
+
"single_word": false,
|
| 139 |
+
"special": false
|
| 140 |
+
},
|
| 141 |
+
"151660": {
|
| 142 |
+
"content": "<|fim_middle|>",
|
| 143 |
+
"lstrip": false,
|
| 144 |
+
"normalized": false,
|
| 145 |
+
"rstrip": false,
|
| 146 |
+
"single_word": false,
|
| 147 |
+
"special": false
|
| 148 |
+
},
|
| 149 |
+
"151661": {
|
| 150 |
+
"content": "<|fim_suffix|>",
|
| 151 |
+
"lstrip": false,
|
| 152 |
+
"normalized": false,
|
| 153 |
+
"rstrip": false,
|
| 154 |
+
"single_word": false,
|
| 155 |
+
"special": false
|
| 156 |
+
},
|
| 157 |
+
"151662": {
|
| 158 |
+
"content": "<|fim_pad|>",
|
| 159 |
+
"lstrip": false,
|
| 160 |
+
"normalized": false,
|
| 161 |
+
"rstrip": false,
|
| 162 |
+
"single_word": false,
|
| 163 |
+
"special": false
|
| 164 |
+
},
|
| 165 |
+
"151663": {
|
| 166 |
+
"content": "<|repo_name|>",
|
| 167 |
+
"lstrip": false,
|
| 168 |
+
"normalized": false,
|
| 169 |
+
"rstrip": false,
|
| 170 |
+
"single_word": false,
|
| 171 |
+
"special": false
|
| 172 |
+
},
|
| 173 |
+
"151664": {
|
| 174 |
+
"content": "<|file_sep|>",
|
| 175 |
+
"lstrip": false,
|
| 176 |
+
"normalized": false,
|
| 177 |
+
"rstrip": false,
|
| 178 |
+
"single_word": false,
|
| 179 |
+
"special": false
|
| 180 |
+
}
|
| 181 |
+
},
|
| 182 |
+
"additional_special_tokens": [
|
| 183 |
+
"<|im_start|>",
|
| 184 |
+
"<|im_end|>",
|
| 185 |
+
"<|object_ref_start|>",
|
| 186 |
+
"<|object_ref_end|>",
|
| 187 |
+
"<|box_start|>",
|
| 188 |
+
"<|box_end|>",
|
| 189 |
+
"<|quad_start|>",
|
| 190 |
+
"<|quad_end|>",
|
| 191 |
+
"<|vision_start|>",
|
| 192 |
+
"<|vision_end|>",
|
| 193 |
+
"<|vision_pad|>",
|
| 194 |
+
"<|image_pad|>",
|
| 195 |
+
"<|video_pad|>"
|
| 196 |
+
],
|
| 197 |
+
"bos_token": null,
|
| 198 |
+
"clean_up_tokenization_spaces": false,
|
| 199 |
+
"eos_token": "<|endoftext|>",
|
| 200 |
+
"errors": "replace",
|
| 201 |
+
"extra_special_tokens": {},
|
| 202 |
+
"model_max_length": 32768,
|
| 203 |
+
"pad_token": "<|endoftext|>",
|
| 204 |
+
"split_special_tokens": false,
|
| 205 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 206 |
+
"unk_token": null
|
| 207 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/trainer_state.json
ADDED
|
@@ -0,0 +1,1134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02002002002002002,
|
| 14 |
+
"grad_norm": 0.05460292845964432,
|
| 15 |
+
"grpo_mean_advantage": -1.3560057254835556e-07,
|
| 16 |
+
"grpo_mean_group_score": 0.5922331809997559,
|
| 17 |
+
"grpo_mean_kl_div": 0.0,
|
| 18 |
+
"grpo_std_advantage": 3.0318567496578908e-06,
|
| 19 |
+
"learning_rate": 8.000000000000001e-07,
|
| 20 |
+
"loss": 0.007,
|
| 21 |
+
"step": 5
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.04004004004004004,
|
| 25 |
+
"grad_norm": 0.0679207444190979,
|
| 26 |
+
"grpo_mean_advantage": 3.6619603633880615e-06,
|
| 27 |
+
"grpo_mean_group_score": 0.5561589002609253,
|
| 28 |
+
"grpo_mean_kl_div": 0.0,
|
| 29 |
+
"grpo_std_advantage": 1.6246918676188216e-05,
|
| 30 |
+
"learning_rate": 1.8000000000000001e-06,
|
| 31 |
+
"loss": 0.0107,
|
| 32 |
+
"step": 10
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06006006006006006,
|
| 36 |
+
"grad_norm": 0.05788416787981987,
|
| 37 |
+
"grpo_mean_advantage": -1.0654330395709621e-07,
|
| 38 |
+
"grpo_mean_group_score": 0.5759152173995972,
|
| 39 |
+
"grpo_mean_kl_div": 0.0,
|
| 40 |
+
"grpo_std_advantage": 5.399440965447866e-07,
|
| 41 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 42 |
+
"loss": 0.007,
|
| 43 |
+
"step": 15
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.08008008008008008,
|
| 47 |
+
"grad_norm": 0.0746568813920021,
|
| 48 |
+
"grpo_mean_advantage": -5.871057737749652e-07,
|
| 49 |
+
"grpo_mean_group_score": 0.5127314329147339,
|
| 50 |
+
"grpo_mean_kl_div": 0.0,
|
| 51 |
+
"grpo_std_advantage": 2.6951597646984737e-06,
|
| 52 |
+
"learning_rate": 3.8000000000000005e-06,
|
| 53 |
+
"loss": 0.0246,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.1001001001001001,
|
| 58 |
+
"grad_norm": 0.11442846059799194,
|
| 59 |
+
"grpo_mean_advantage": 6.370246410369873e-07,
|
| 60 |
+
"grpo_mean_group_score": 0.539706826210022,
|
| 61 |
+
"grpo_mean_kl_div": 0.0,
|
| 62 |
+
"grpo_std_advantage": 2.8908377771585947e-06,
|
| 63 |
+
"learning_rate": 4.800000000000001e-06,
|
| 64 |
+
"loss": 0.0337,
|
| 65 |
+
"step": 25
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.12012012012012012,
|
| 69 |
+
"grad_norm": 0.05778791010379791,
|
| 70 |
+
"grpo_mean_advantage": 6.705522359595761e-09,
|
| 71 |
+
"grpo_mean_group_score": 0.5812538862228394,
|
| 72 |
+
"grpo_mean_kl_div": 0.0,
|
| 73 |
+
"grpo_std_advantage": 6.189450800775376e-07,
|
| 74 |
+
"learning_rate": 4.999125183044924e-06,
|
| 75 |
+
"loss": 0.0171,
|
| 76 |
+
"step": 30
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.14014014014014015,
|
| 80 |
+
"grad_norm": 0.05819695070385933,
|
| 81 |
+
"grpo_mean_advantage": 3.859400692363124e-07,
|
| 82 |
+
"grpo_mean_group_score": 0.5909844636917114,
|
| 83 |
+
"grpo_mean_kl_div": 0.0,
|
| 84 |
+
"grpo_std_advantage": 1.6833292875162442e-06,
|
| 85 |
+
"learning_rate": 4.995572288443412e-06,
|
| 86 |
+
"loss": 0.0145,
|
| 87 |
+
"step": 35
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.16016016016016016,
|
| 91 |
+
"grad_norm": 0.07968433201313019,
|
| 92 |
+
"grpo_mean_advantage": 2.600252742013254e-07,
|
| 93 |
+
"grpo_mean_group_score": 0.5630953907966614,
|
| 94 |
+
"grpo_mean_kl_div": 0.0,
|
| 95 |
+
"grpo_std_advantage": 1.4095899132371414e-06,
|
| 96 |
+
"learning_rate": 4.98929052218411e-06,
|
| 97 |
+
"loss": 0.0196,
|
| 98 |
+
"step": 40
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.18018018018018017,
|
| 102 |
+
"grad_norm": 0.0733402892947197,
|
| 103 |
+
"grpo_mean_advantage": -1.2591480924584175e-07,
|
| 104 |
+
"grpo_mean_group_score": 0.5604403614997864,
|
| 105 |
+
"grpo_mean_kl_div": 0.0,
|
| 106 |
+
"grpo_std_advantage": 1.0309080380466185e-06,
|
| 107 |
+
"learning_rate": 4.980286753286196e-06,
|
| 108 |
+
"loss": 0.0186,
|
| 109 |
+
"step": 45
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.2002002002002002,
|
| 113 |
+
"grad_norm": 0.07136482000350952,
|
| 114 |
+
"grpo_mean_advantage": -2.808868941883702e-07,
|
| 115 |
+
"grpo_mean_group_score": 0.5971035957336426,
|
| 116 |
+
"grpo_mean_kl_div": 0.0,
|
| 117 |
+
"grpo_std_advantage": 1.5696078889959608e-06,
|
| 118 |
+
"learning_rate": 4.9685708272387645e-06,
|
| 119 |
+
"loss": 0.0286,
|
| 120 |
+
"step": 50
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.22022022022022023,
|
| 124 |
+
"grad_norm": 0.08851475268602371,
|
| 125 |
+
"grpo_mean_advantage": 2.6822089438383045e-08,
|
| 126 |
+
"grpo_mean_group_score": 0.5892971754074097,
|
| 127 |
+
"grpo_mean_kl_div": 0.0,
|
| 128 |
+
"grpo_std_advantage": 3.7878271541558206e-07,
|
| 129 |
+
"learning_rate": 4.9541555552349404e-06,
|
| 130 |
+
"loss": 0.0054,
|
| 131 |
+
"step": 55
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.24024024024024024,
|
| 135 |
+
"grad_norm": 0.07778509706258774,
|
| 136 |
+
"grpo_mean_advantage": -5.662441182607836e-08,
|
| 137 |
+
"grpo_mean_group_score": 0.564322292804718,
|
| 138 |
+
"grpo_mean_kl_div": 0.0,
|
| 139 |
+
"grpo_std_advantage": 6.128998393251095e-07,
|
| 140 |
+
"learning_rate": 4.9370567001630155e-06,
|
| 141 |
+
"loss": -0.0074,
|
| 142 |
+
"step": 60
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2602602602602603,
|
| 146 |
+
"grad_norm": 0.08740051090717316,
|
| 147 |
+
"grpo_mean_advantage": -1.5944242193199898e-07,
|
| 148 |
+
"grpo_mean_group_score": 0.562497615814209,
|
| 149 |
+
"grpo_mean_kl_div": 0.0,
|
| 150 |
+
"grpo_std_advantage": 1.6374274309782777e-06,
|
| 151 |
+
"learning_rate": 4.917292959369968e-06,
|
| 152 |
+
"loss": 0.0145,
|
| 153 |
+
"step": 65
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.2802802802802803,
|
| 157 |
+
"grad_norm": 0.19070060551166534,
|
| 158 |
+
"grpo_mean_advantage": 1.6838312433264946e-07,
|
| 159 |
+
"grpo_mean_group_score": 0.5904761552810669,
|
| 160 |
+
"grpo_mean_kl_div": 0.0,
|
| 161 |
+
"grpo_std_advantage": 8.536571272088622e-07,
|
| 162 |
+
"learning_rate": 4.8948859442161876e-06,
|
| 163 |
+
"loss": 0.0257,
|
| 164 |
+
"step": 70
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3003003003003003,
|
| 168 |
+
"grad_norm": 0.07321271300315857,
|
| 169 |
+
"grpo_mean_advantage": 1.1175870895385742e-07,
|
| 170 |
+
"grpo_mean_group_score": 0.5765624046325684,
|
| 171 |
+
"grpo_mean_kl_div": 0.0,
|
| 172 |
+
"grpo_std_advantage": 6.451961667153228e-07,
|
| 173 |
+
"learning_rate": 4.869860156443768e-06,
|
| 174 |
+
"loss": 0.0024,
|
| 175 |
+
"step": 75
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.3203203203203203,
|
| 179 |
+
"grad_norm": 0.07126748561859131,
|
| 180 |
+
"grpo_mean_advantage": -1.4603138254187797e-07,
|
| 181 |
+
"grpo_mean_group_score": 0.5858271718025208,
|
| 182 |
+
"grpo_mean_kl_div": 0.0,
|
| 183 |
+
"grpo_std_advantage": 1.1309343790344428e-06,
|
| 184 |
+
"learning_rate": 4.842242961384211e-06,
|
| 185 |
+
"loss": 0.0277,
|
| 186 |
+
"step": 80
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.34034034034034033,
|
| 190 |
+
"grad_norm": 0.08629189431667328,
|
| 191 |
+
"grpo_mean_advantage": -1.817941665649414e-06,
|
| 192 |
+
"grpo_mean_group_score": 0.5871662497520447,
|
| 193 |
+
"grpo_mean_kl_div": 0.0,
|
| 194 |
+
"grpo_std_advantage": 1.1141768482048064e-05,
|
| 195 |
+
"learning_rate": 4.812064558034847e-06,
|
| 196 |
+
"loss": 0.0246,
|
| 197 |
+
"step": 85
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.36036036036036034,
|
| 201 |
+
"grad_norm": 0.0998779758810997,
|
| 202 |
+
"grpo_mean_advantage": 1.8179416372277046e-07,
|
| 203 |
+
"grpo_mean_group_score": 0.5330992937088013,
|
| 204 |
+
"grpo_mean_kl_div": 0.0,
|
| 205 |
+
"grpo_std_advantage": 6.210335072864837e-07,
|
| 206 |
+
"learning_rate": 4.779357946036662e-06,
|
| 207 |
+
"loss": 0.0056,
|
| 208 |
+
"step": 90
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.38038038038038036,
|
| 212 |
+
"grad_norm": 0.10614689439535141,
|
| 213 |
+
"grpo_mean_advantage": -2.972781771859445e-07,
|
| 214 |
+
"grpo_mean_group_score": 0.5265295505523682,
|
| 215 |
+
"grpo_mean_kl_div": 0.0,
|
| 216 |
+
"grpo_std_advantage": 3.1582342217006953e-06,
|
| 217 |
+
"learning_rate": 4.74415888958968e-06,
|
| 218 |
+
"loss": 0.0053,
|
| 219 |
+
"step": 95
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.4004004004004004,
|
| 223 |
+
"grad_norm": 0.10345634073019028,
|
| 224 |
+
"grpo_mean_advantage": -7.033348197182931e-07,
|
| 225 |
+
"grpo_mean_group_score": 0.5660771131515503,
|
| 226 |
+
"grpo_mean_kl_div": 0.0,
|
| 227 |
+
"grpo_std_advantage": 4.245831405569334e-06,
|
| 228 |
+
"learning_rate": 4.706505878345343e-06,
|
| 229 |
+
"loss": 0.0134,
|
| 230 |
+
"step": 100
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.42042042042042044,
|
| 234 |
+
"grad_norm": 0.10077933222055435,
|
| 235 |
+
"grpo_mean_advantage": 1.1920928955078125e-07,
|
| 236 |
+
"grpo_mean_group_score": 0.57631915807724,
|
| 237 |
+
"grpo_mean_kl_div": 0.0,
|
| 238 |
+
"grpo_std_advantage": 3.2809634831210133e-07,
|
| 239 |
+
"learning_rate": 4.666440085318626e-06,
|
| 240 |
+
"loss": 0.0004,
|
| 241 |
+
"step": 105
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.44044044044044045,
|
| 245 |
+
"grad_norm": 0.09548182785511017,
|
| 246 |
+
"grpo_mean_advantage": -4.0978193283081055e-07,
|
| 247 |
+
"grpo_mean_group_score": 0.546563982963562,
|
| 248 |
+
"grpo_mean_kl_div": 0.0,
|
| 249 |
+
"grpo_std_advantage": 6.0397578636184335e-06,
|
| 250 |
+
"learning_rate": 4.624005321865968e-06,
|
| 251 |
+
"loss": 0.0033,
|
| 252 |
+
"step": 110
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.46046046046046046,
|
| 256 |
+
"grad_norm": 0.09417816251516342,
|
| 257 |
+
"grpo_mean_advantage": -1.467764434437413e-07,
|
| 258 |
+
"grpo_mean_group_score": 0.5519219636917114,
|
| 259 |
+
"grpo_mean_kl_div": 0.0,
|
| 260 |
+
"grpo_std_advantage": 2.2689375782647403e-06,
|
| 261 |
+
"learning_rate": 4.57924798977818e-06,
|
| 262 |
+
"loss": 0.0095,
|
| 263 |
+
"step": 115
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.4804804804804805,
|
| 267 |
+
"grad_norm": 0.10022275149822235,
|
| 268 |
+
"grpo_mean_advantage": -5.215406329028838e-09,
|
| 269 |
+
"grpo_mean_group_score": 0.5490407943725586,
|
| 270 |
+
"grpo_mean_kl_div": 0.0,
|
| 271 |
+
"grpo_std_advantage": 7.929010621410271e-07,
|
| 272 |
+
"learning_rate": 4.532217030540781e-06,
|
| 273 |
+
"loss": 0.0006,
|
| 274 |
+
"step": 120
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.5005005005005005,
|
| 278 |
+
"grad_norm": 0.14057794213294983,
|
| 279 |
+
"grpo_mean_advantage": -5.7369469175228005e-08,
|
| 280 |
+
"grpo_mean_group_score": 0.5646580457687378,
|
| 281 |
+
"grpo_mean_kl_div": 0.0,
|
| 282 |
+
"grpo_std_advantage": 1.2823379620385822e-06,
|
| 283 |
+
"learning_rate": 4.482963871817195e-06,
|
| 284 |
+
"loss": -0.0046,
|
| 285 |
+
"step": 125
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.5205205205205206,
|
| 289 |
+
"grad_norm": 0.12420658767223358,
|
| 290 |
+
"grpo_mean_advantage": 2.9876827056796174e-07,
|
| 291 |
+
"grpo_mean_group_score": 0.6111599802970886,
|
| 292 |
+
"grpo_mean_kl_div": 0.0,
|
| 293 |
+
"grpo_std_advantage": 1.0496698905626545e-06,
|
| 294 |
+
"learning_rate": 4.4315423712133595e-06,
|
| 295 |
+
"loss": -0.003,
|
| 296 |
+
"step": 130
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.5405405405405406,
|
| 300 |
+
"grad_norm": 0.14342808723449707,
|
| 301 |
+
"grpo_mean_advantage": 1.5869736103013565e-07,
|
| 302 |
+
"grpo_mean_group_score": 0.5619662404060364,
|
| 303 |
+
"grpo_mean_kl_div": 0.0,
|
| 304 |
+
"grpo_std_advantage": 1.2748531617035042e-06,
|
| 305 |
+
"learning_rate": 4.378008757385222e-06,
|
| 306 |
+
"loss": 0.0154,
|
| 307 |
+
"step": 135
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.5605605605605606,
|
| 311 |
+
"grad_norm": 0.14729444682598114,
|
| 312 |
+
"grpo_mean_advantage": 3.0100346748440643e-07,
|
| 313 |
+
"grpo_mean_group_score": 0.5795454978942871,
|
| 314 |
+
"grpo_mean_kl_div": 0.0,
|
| 315 |
+
"grpo_std_advantage": 2.4499684059264837e-06,
|
| 316 |
+
"learning_rate": 4.322421568553529e-06,
|
| 317 |
+
"loss": -0.0262,
|
| 318 |
+
"step": 140
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.5805805805805806,
|
| 322 |
+
"grad_norm": 0.15249410271644592,
|
| 323 |
+
"grpo_mean_advantage": -3.233552092751779e-07,
|
| 324 |
+
"grpo_mean_group_score": 0.5804953575134277,
|
| 325 |
+
"grpo_mean_kl_div": 0.0,
|
| 326 |
+
"grpo_std_advantage": 1.248456669600273e-06,
|
| 327 |
+
"learning_rate": 4.2648415884931476e-06,
|
| 328 |
+
"loss": 0.0018,
|
| 329 |
+
"step": 145
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.6006006006006006,
|
| 333 |
+
"grad_norm": 0.1841023564338684,
|
| 334 |
+
"grpo_mean_advantage": 3.2261013416245987e-07,
|
| 335 |
+
"grpo_mean_group_score": 0.5628539323806763,
|
| 336 |
+
"grpo_mean_kl_div": 0.0,
|
| 337 |
+
"grpo_std_advantage": 1.4773489738217904e-06,
|
| 338 |
+
"learning_rate": 4.205331780066892e-06,
|
| 339 |
+
"loss": -0.017,
|
| 340 |
+
"step": 150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6206206206206206,
|
| 344 |
+
"grad_norm": 0.18597163259983063,
|
| 345 |
+
"grpo_mean_advantage": -2.5331974029541016e-07,
|
| 346 |
+
"grpo_mean_group_score": 0.5727725625038147,
|
| 347 |
+
"grpo_mean_kl_div": 0.0,
|
| 348 |
+
"grpo_std_advantage": 1.5092309695319273e-06,
|
| 349 |
+
"learning_rate": 4.1439572163765615e-06,
|
| 350 |
+
"loss": 0.0044,
|
| 351 |
+
"step": 155
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.6406406406406406,
|
| 355 |
+
"grad_norm": 0.18310388922691345,
|
| 356 |
+
"grpo_mean_advantage": -6.780028627417778e-08,
|
| 357 |
+
"grpo_mean_group_score": 0.5833909511566162,
|
| 358 |
+
"grpo_mean_kl_div": 0.0,
|
| 359 |
+
"grpo_std_advantage": 8.550978805033083e-07,
|
| 360 |
+
"learning_rate": 4.0807850096064605e-06,
|
| 361 |
+
"loss": -0.005,
|
| 362 |
+
"step": 160
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6606606606606606,
|
| 366 |
+
"grad_norm": 0.2192923128604889,
|
| 367 |
+
"grpo_mean_advantage": -5.587935447692871e-08,
|
| 368 |
+
"grpo_mean_group_score": 0.5742615461349487,
|
| 369 |
+
"grpo_mean_kl_div": 0.0,
|
| 370 |
+
"grpo_std_advantage": 3.564579174053506e-07,
|
| 371 |
+
"learning_rate": 4.015884237637206e-06,
|
| 372 |
+
"loss": -0.015,
|
| 373 |
+
"step": 165
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.6806806806806807,
|
| 377 |
+
"grad_norm": 0.16708803176879883,
|
| 378 |
+
"grpo_mean_advantage": -5.327165126800537e-07,
|
| 379 |
+
"grpo_mean_group_score": 0.5758188962936401,
|
| 380 |
+
"grpo_mean_kl_div": 0.0,
|
| 381 |
+
"grpo_std_advantage": 2.309018327650847e-06,
|
| 382 |
+
"learning_rate": 3.949325868510083e-06,
|
| 383 |
+
"loss": -0.0314,
|
| 384 |
+
"step": 170
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7007007007007007,
|
| 388 |
+
"grad_norm": 0.3401262164115906,
|
| 389 |
+
"grpo_mean_advantage": 5.863606702405377e-07,
|
| 390 |
+
"grpo_mean_group_score": 0.5767683982849121,
|
| 391 |
+
"grpo_mean_kl_div": 0.0,
|
| 392 |
+
"grpo_std_advantage": 2.4449204829579685e-06,
|
| 393 |
+
"learning_rate": 3.881182682824534e-06,
|
| 394 |
+
"loss": -0.0441,
|
| 395 |
+
"step": 175
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.7207207207207207,
|
| 399 |
+
"grad_norm": 0.1931898146867752,
|
| 400 |
+
"grpo_mean_advantage": 3.2186508747145126e-07,
|
| 401 |
+
"grpo_mean_group_score": 0.586772084236145,
|
| 402 |
+
"grpo_mean_kl_div": 0.0,
|
| 403 |
+
"grpo_std_advantage": 2.293551688126172e-06,
|
| 404 |
+
"learning_rate": 3.811529194153635e-06,
|
| 405 |
+
"loss": -0.0162,
|
| 406 |
+
"step": 180
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7407407407407407,
|
| 410 |
+
"grad_norm": 0.2537969648838043,
|
| 411 |
+
"grpo_mean_advantage": -4.470348358154297e-08,
|
| 412 |
+
"grpo_mean_group_score": 0.549396276473999,
|
| 413 |
+
"grpo_mean_kl_div": 0.0,
|
| 414 |
+
"grpo_std_advantage": 3.7067667335577426e-07,
|
| 415 |
+
"learning_rate": 3.7404415675646054e-06,
|
| 416 |
+
"loss": -0.0386,
|
| 417 |
+
"step": 185
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.7607607607607607,
|
| 421 |
+
"grad_norm": 0.20326584577560425,
|
| 422 |
+
"grpo_mean_advantage": -2.1010637851759384e-07,
|
| 423 |
+
"grpo_mean_group_score": 0.5798425078392029,
|
| 424 |
+
"grpo_mean_kl_div": 0.0,
|
| 425 |
+
"grpo_std_advantage": 1.1695076409523608e-06,
|
| 426 |
+
"learning_rate": 3.667997536333424e-06,
|
| 427 |
+
"loss": -0.037,
|
| 428 |
+
"step": 190
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.7807807807807807,
|
| 432 |
+
"grad_norm": 0.25048357248306274,
|
| 433 |
+
"grpo_mean_advantage": 1.765787658314366e-07,
|
| 434 |
+
"grpo_mean_group_score": 0.5584167838096619,
|
| 435 |
+
"grpo_mean_kl_div": 0.0,
|
| 436 |
+
"grpo_std_advantage": 2.429934738756856e-06,
|
| 437 |
+
"learning_rate": 3.59427631694463e-06,
|
| 438 |
+
"loss": -0.0292,
|
| 439 |
+
"step": 195
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.8008008008008008,
|
| 443 |
+
"grad_norm": 0.2687569260597229,
|
| 444 |
+
"grpo_mean_advantage": 1.6540289493605087e-07,
|
| 445 |
+
"grpo_mean_group_score": 0.5676193237304688,
|
| 446 |
+
"grpo_mean_kl_div": 0.0,
|
| 447 |
+
"grpo_std_advantage": 2.6342788714828203e-06,
|
| 448 |
+
"learning_rate": 3.5193585224692595e-06,
|
| 449 |
+
"loss": -0.0454,
|
| 450 |
+
"step": 200
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8208208208208209,
|
| 454 |
+
"grad_norm": 0.22301620244979858,
|
| 455 |
+
"grpo_mean_advantage": -1.0944902442133753e-06,
|
| 456 |
+
"grpo_mean_group_score": 0.5669739842414856,
|
| 457 |
+
"grpo_mean_kl_div": 0.0,
|
| 458 |
+
"grpo_std_advantage": 5.346942998585291e-06,
|
| 459 |
+
"learning_rate": 3.44332607441564e-06,
|
| 460 |
+
"loss": -0.0423,
|
| 461 |
+
"step": 205
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.8408408408408409,
|
| 465 |
+
"grad_norm": 0.3040211498737335,
|
| 466 |
+
"grpo_mean_advantage": 2.4065374759629776e-07,
|
| 467 |
+
"grpo_mean_group_score": 0.5922158360481262,
|
| 468 |
+
"grpo_mean_kl_div": 0.0,
|
| 469 |
+
"grpo_std_advantage": 1.6327536513927043e-06,
|
| 470 |
+
"learning_rate": 3.3662621131494204e-06,
|
| 471 |
+
"loss": -0.0857,
|
| 472 |
+
"step": 210
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.8608608608608609,
|
| 476 |
+
"grad_norm": 0.27231141924858093,
|
| 477 |
+
"grpo_mean_advantage": -5.21540641784668e-08,
|
| 478 |
+
"grpo_mean_group_score": 0.5473950505256653,
|
| 479 |
+
"grpo_mean_kl_div": 0.0,
|
| 480 |
+
"grpo_std_advantage": 5.847922466273303e-07,
|
| 481 |
+
"learning_rate": 3.2882509069808044e-06,
|
| 482 |
+
"loss": -0.0278,
|
| 483 |
+
"step": 215
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.8808808808808809,
|
| 487 |
+
"grad_norm": 0.3571636378765106,
|
| 488 |
+
"grpo_mean_advantage": 6.541609991472797e-07,
|
| 489 |
+
"grpo_mean_group_score": 0.5880032777786255,
|
| 490 |
+
"grpo_mean_kl_div": 0.0,
|
| 491 |
+
"grpo_std_advantage": 4.072162937518442e-06,
|
| 492 |
+
"learning_rate": 3.2093777600183873e-06,
|
| 493 |
+
"loss": -0.0727,
|
| 494 |
+
"step": 220
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9009009009009009,
|
| 498 |
+
"grad_norm": 0.306273490190506,
|
| 499 |
+
"grpo_mean_advantage": -1.2218951894737984e-07,
|
| 500 |
+
"grpo_mean_group_score": 0.5835092663764954,
|
| 501 |
+
"grpo_mean_kl_div": 0.0,
|
| 502 |
+
"grpo_std_advantage": 4.386006935419573e-07,
|
| 503 |
+
"learning_rate": 3.1297289188903705e-06,
|
| 504 |
+
"loss": -0.0464,
|
| 505 |
+
"step": 225
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.9209209209209209,
|
| 509 |
+
"grad_norm": 0.2700377106666565,
|
| 510 |
+
"grpo_mean_advantage": 1.7605722177904681e-06,
|
| 511 |
+
"grpo_mean_group_score": 0.5394966006278992,
|
| 512 |
+
"grpo_mean_kl_div": 0.0,
|
| 513 |
+
"grpo_std_advantage": 8.007580618141219e-06,
|
| 514 |
+
"learning_rate": 3.049391478435133e-06,
|
| 515 |
+
"loss": -0.0295,
|
| 516 |
+
"step": 230
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9409409409409409,
|
| 520 |
+
"grad_norm": 0.39531761407852173,
|
| 521 |
+
"grpo_mean_advantage": -3.3080578987210174e-07,
|
| 522 |
+
"grpo_mean_group_score": 0.5687432289123535,
|
| 523 |
+
"grpo_mean_kl_div": 0.0,
|
| 524 |
+
"grpo_std_advantage": 1.551636614749441e-06,
|
| 525 |
+
"learning_rate": 2.9684532864643123e-06,
|
| 526 |
+
"loss": -0.031,
|
| 527 |
+
"step": 235
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.960960960960961,
|
| 531 |
+
"grad_norm": 0.5987040996551514,
|
| 532 |
+
"grpo_mean_advantage": 2.712011450967111e-07,
|
| 533 |
+
"grpo_mean_group_score": 0.5550583600997925,
|
| 534 |
+
"grpo_mean_kl_div": 0.0,
|
| 535 |
+
"grpo_std_advantage": 1.4400844747797237e-06,
|
| 536 |
+
"learning_rate": 2.887002847702504e-06,
|
| 537 |
+
"loss": -0.0789,
|
| 538 |
+
"step": 240
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.980980980980981,
|
| 542 |
+
"grad_norm": 0.5680716037750244,
|
| 543 |
+
"grpo_mean_advantage": -3.2857059295565705e-07,
|
| 544 |
+
"grpo_mean_group_score": 0.558111310005188,
|
| 545 |
+
"grpo_mean_kl_div": 0.0,
|
| 546 |
+
"grpo_std_advantage": 2.105091425619321e-06,
|
| 547 |
+
"learning_rate": 2.8051292270086506e-06,
|
| 548 |
+
"loss": -0.1131,
|
| 549 |
+
"step": 245
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.0,
|
| 553 |
+
"grad_norm": 0.6204046010971069,
|
| 554 |
+
"grpo_mean_advantage": 4.470348358154297e-08,
|
| 555 |
+
"grpo_mean_group_score": 0.6196198463439941,
|
| 556 |
+
"grpo_mean_kl_div": 0.0,
|
| 557 |
+
"grpo_std_advantage": 5.315724820320611e-07,
|
| 558 |
+
"learning_rate": 2.722921951984927e-06,
|
| 559 |
+
"loss": -0.2232,
|
| 560 |
+
"step": 250
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.02002002002002,
|
| 564 |
+
"grad_norm": 0.8389026522636414,
|
| 565 |
+
"grpo_mean_advantage": 9.290873776990338e-07,
|
| 566 |
+
"grpo_mean_group_score": 0.582168459892273,
|
| 567 |
+
"grpo_mean_kl_div": 0.0,
|
| 568 |
+
"grpo_std_advantage": 4.219644324621186e-06,
|
| 569 |
+
"learning_rate": 2.640470915079614e-06,
|
| 570 |
+
"loss": -0.1363,
|
| 571 |
+
"step": 255
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.04004004004004,
|
| 575 |
+
"grad_norm": 0.9067686796188354,
|
| 576 |
+
"grpo_mean_advantage": 2.533197474008375e-08,
|
| 577 |
+
"grpo_mean_group_score": 0.5551307797431946,
|
| 578 |
+
"grpo_mean_kl_div": 0.0,
|
| 579 |
+
"grpo_std_advantage": 1.6600588992332632e-07,
|
| 580 |
+
"learning_rate": 2.557866275291035e-06,
|
| 581 |
+
"loss": -0.1868,
|
| 582 |
+
"step": 260
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.06006006006006,
|
| 586 |
+
"grad_norm": 0.9277902841567993,
|
| 587 |
+
"grpo_mean_advantage": -5.662441182607836e-08,
|
| 588 |
+
"grpo_mean_group_score": 0.535040020942688,
|
| 589 |
+
"grpo_mean_kl_div": 0.0,
|
| 590 |
+
"grpo_std_advantage": 1.0909400316450046e-06,
|
| 591 |
+
"learning_rate": 2.4751983595800093e-06,
|
| 592 |
+
"loss": -0.1792,
|
| 593 |
+
"step": 265
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.08008008008008,
|
| 597 |
+
"grad_norm": 1.0715463161468506,
|
| 598 |
+
"grpo_mean_advantage": -9.536743306171047e-08,
|
| 599 |
+
"grpo_mean_group_score": 0.5673571825027466,
|
| 600 |
+
"grpo_mean_kl_div": 0.0,
|
| 601 |
+
"grpo_std_advantage": 5.838213610331877e-07,
|
| 602 |
+
"learning_rate": 2.392557564098649e-06,
|
| 603 |
+
"loss": -0.1691,
|
| 604 |
+
"step": 270
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.1001001001001,
|
| 608 |
+
"grad_norm": 0.7759184837341309,
|
| 609 |
+
"grpo_mean_advantage": 3.278255533700758e-08,
|
| 610 |
+
"grpo_mean_group_score": 0.5874732732772827,
|
| 611 |
+
"grpo_mean_kl_div": 0.0,
|
| 612 |
+
"grpo_std_advantage": 9.317170679423725e-07,
|
| 613 |
+
"learning_rate": 2.3100342553434924e-06,
|
| 614 |
+
"loss": -0.1655,
|
| 615 |
+
"step": 275
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.12012012012012,
|
| 619 |
+
"grad_norm": 0.9387398958206177,
|
| 620 |
+
"grpo_mean_advantage": -1.206994113545079e-07,
|
| 621 |
+
"grpo_mean_group_score": 0.5569106340408325,
|
| 622 |
+
"grpo_mean_kl_div": 0.0,
|
| 623 |
+
"grpo_std_advantage": 6.201085511747806e-07,
|
| 624 |
+
"learning_rate": 2.2277186713410688e-06,
|
| 625 |
+
"loss": -0.1821,
|
| 626 |
+
"step": 280
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.14014014014014,
|
| 630 |
+
"grad_norm": 1.6132302284240723,
|
| 631 |
+
"grpo_mean_advantage": 4.470348358154297e-08,
|
| 632 |
+
"grpo_mean_group_score": 0.5578873157501221,
|
| 633 |
+
"grpo_mean_kl_div": 0.0,
|
| 634 |
+
"grpo_std_advantage": 6.115651558502577e-07,
|
| 635 |
+
"learning_rate": 2.1457008229739395e-06,
|
| 636 |
+
"loss": -0.2102,
|
| 637 |
+
"step": 285
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.16016016016016,
|
| 641 |
+
"grad_norm": 0.8679026961326599,
|
| 642 |
+
"grpo_mean_advantage": -3.3453108017056365e-07,
|
| 643 |
+
"grpo_mean_group_score": 0.5735999345779419,
|
| 644 |
+
"grpo_mean_kl_div": 0.0,
|
| 645 |
+
"grpo_std_advantage": 3.5326345368957845e-06,
|
| 646 |
+
"learning_rate": 2.0640703955551214e-06,
|
| 647 |
+
"loss": -0.2937,
|
| 648 |
+
"step": 290
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.1801801801801801,
|
| 652 |
+
"grad_norm": 1.0550166368484497,
|
| 653 |
+
"grpo_mean_advantage": -1.110136480519941e-07,
|
| 654 |
+
"grpo_mean_group_score": 0.5626259446144104,
|
| 655 |
+
"grpo_mean_kl_div": 0.0,
|
| 656 |
+
"grpo_std_advantage": 4.731904823529476e-07,
|
| 657 |
+
"learning_rate": 1.9829166507585084e-06,
|
| 658 |
+
"loss": -0.2598,
|
| 659 |
+
"step": 295
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 1.2002002002002001,
|
| 663 |
+
"grad_norm": 1.2819372415542603,
|
| 664 |
+
"grpo_mean_advantage": -5.08874677507265e-07,
|
| 665 |
+
"grpo_mean_group_score": 0.5463050603866577,
|
| 666 |
+
"grpo_mean_kl_div": 0.0,
|
| 667 |
+
"grpo_std_advantage": 1.840126174101897e-06,
|
| 668 |
+
"learning_rate": 1.90232832901255e-06,
|
| 669 |
+
"loss": -0.2546,
|
| 670 |
+
"step": 300
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.2202202202202201,
|
| 674 |
+
"grad_norm": 1.0188143253326416,
|
| 675 |
+
"grpo_mean_advantage": 1.01327898960335e-07,
|
| 676 |
+
"grpo_mean_group_score": 0.5352144241333008,
|
| 677 |
+
"grpo_mean_kl_div": 0.0,
|
| 678 |
+
"grpo_std_advantage": 7.798533943059738e-07,
|
| 679 |
+
"learning_rate": 1.82239355246389e-06,
|
| 680 |
+
"loss": -0.1809,
|
| 681 |
+
"step": 305
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 1.2402402402402402,
|
| 685 |
+
"grad_norm": 2.0709052085876465,
|
| 686 |
+
"grpo_mean_advantage": 1.341104507446289e-07,
|
| 687 |
+
"grpo_mean_group_score": 0.5547868013381958,
|
| 688 |
+
"grpo_mean_kl_div": 0.0,
|
| 689 |
+
"grpo_std_advantage": 7.821902840987605e-07,
|
| 690 |
+
"learning_rate": 1.7431997286170923e-06,
|
| 691 |
+
"loss": -0.3559,
|
| 692 |
+
"step": 310
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.2602602602602602,
|
| 696 |
+
"grad_norm": 1.8516215085983276,
|
| 697 |
+
"grpo_mean_advantage": 9.015202806494926e-08,
|
| 698 |
+
"grpo_mean_group_score": 0.5859472751617432,
|
| 699 |
+
"grpo_mean_kl_div": 0.0,
|
| 700 |
+
"grpo_std_advantage": 1.0693488547985908e-06,
|
| 701 |
+
"learning_rate": 1.6648334547558227e-06,
|
| 702 |
+
"loss": -0.3874,
|
| 703 |
+
"step": 315
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.2802802802802802,
|
| 707 |
+
"grad_norm": 1.283104419708252,
|
| 708 |
+
"grpo_mean_advantage": -2.443790378947597e-07,
|
| 709 |
+
"grpo_mean_group_score": 0.5751550793647766,
|
| 710 |
+
"grpo_mean_kl_div": 0.0,
|
| 711 |
+
"grpo_std_advantage": 1.183122208203713e-06,
|
| 712 |
+
"learning_rate": 1.5873804232499862e-06,
|
| 713 |
+
"loss": -0.3467,
|
| 714 |
+
"step": 320
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3003003003003002,
|
| 718 |
+
"grad_norm": 1.4108576774597168,
|
| 719 |
+
"grpo_mean_advantage": -6.705522537231445e-08,
|
| 720 |
+
"grpo_mean_group_score": 0.5497723817825317,
|
| 721 |
+
"grpo_mean_kl_div": 0.0,
|
| 722 |
+
"grpo_std_advantage": 6.109748937888071e-07,
|
| 723 |
+
"learning_rate": 1.51092532785238e-06,
|
| 724 |
+
"loss": -0.1703,
|
| 725 |
+
"step": 325
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.3203203203203202,
|
| 729 |
+
"grad_norm": 1.0421361923217773,
|
| 730 |
+
"grpo_mean_advantage": -1.639127766850379e-08,
|
| 731 |
+
"grpo_mean_group_score": 0.55989670753479,
|
| 732 |
+
"grpo_mean_kl_div": 0.0,
|
| 733 |
+
"grpo_std_advantage": 5.529495297196263e-07,
|
| 734 |
+
"learning_rate": 1.4355517710873184e-06,
|
| 735 |
+
"loss": -0.2918,
|
| 736 |
+
"step": 330
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.3403403403403402,
|
| 740 |
+
"grad_norm": 1.3465828895568848,
|
| 741 |
+
"grpo_mean_advantage": 4.418194237132411e-07,
|
| 742 |
+
"grpo_mean_group_score": 0.5809233784675598,
|
| 743 |
+
"grpo_mean_kl_div": 0.0,
|
| 744 |
+
"grpo_std_advantage": 2.9275292945385445e-06,
|
| 745 |
+
"learning_rate": 1.361342172832502e-06,
|
| 746 |
+
"loss": -0.3069,
|
| 747 |
+
"step": 335
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.3603603603603602,
|
| 751 |
+
"grad_norm": 1.1959459781646729,
|
| 752 |
+
"grpo_mean_advantage": 9.685754776000977e-08,
|
| 753 |
+
"grpo_mean_group_score": 0.5568087100982666,
|
| 754 |
+
"grpo_mean_kl_div": 0.0,
|
| 755 |
+
"grpo_std_advantage": 3.754235251562932e-07,
|
| 756 |
+
"learning_rate": 1.2883776801940884e-06,
|
| 757 |
+
"loss": -0.5594,
|
| 758 |
+
"step": 340
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.3803803803803802,
|
| 762 |
+
"grad_norm": 1.8967422246932983,
|
| 763 |
+
"grpo_mean_advantage": -2.384185791015625e-07,
|
| 764 |
+
"grpo_mean_group_score": 0.5655568838119507,
|
| 765 |
+
"grpo_mean_kl_div": 0.0,
|
| 766 |
+
"grpo_std_advantage": 6.821086913078034e-07,
|
| 767 |
+
"learning_rate": 1.216738078773522e-06,
|
| 768 |
+
"loss": -0.4102,
|
| 769 |
+
"step": 345
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.4004004004004005,
|
| 773 |
+
"grad_norm": 2.221132755279541,
|
| 774 |
+
"grpo_mean_advantage": -8.717179156292332e-08,
|
| 775 |
+
"grpo_mean_group_score": 0.6089578866958618,
|
| 776 |
+
"grpo_mean_kl_div": 0.0,
|
| 777 |
+
"grpo_std_advantage": 2.500940354366321e-06,
|
| 778 |
+
"learning_rate": 1.146501705423155e-06,
|
| 779 |
+
"loss": -0.338,
|
| 780 |
+
"step": 350
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.4204204204204205,
|
| 784 |
+
"grad_norm": 2.3640377521514893,
|
| 785 |
+
"grpo_mean_advantage": 2.1606683731079102e-07,
|
| 786 |
+
"grpo_mean_group_score": 0.6129671335220337,
|
| 787 |
+
"grpo_mean_kl_div": 0.0,
|
| 788 |
+
"grpo_std_advantage": 1.4568390724889468e-06,
|
| 789 |
+
"learning_rate": 1.0777453625860474e-06,
|
| 790 |
+
"loss": -0.4985,
|
| 791 |
+
"step": 355
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.4404404404404405,
|
| 795 |
+
"grad_norm": 1.9084734916687012,
|
| 796 |
+
"grpo_mean_advantage": -3.725290298461914e-09,
|
| 797 |
+
"grpo_mean_group_score": 0.5562310814857483,
|
| 798 |
+
"grpo_mean_kl_div": 0.0,
|
| 799 |
+
"grpo_std_advantage": 2.965894054796081e-06,
|
| 800 |
+
"learning_rate": 1.0105442343136184e-06,
|
| 801 |
+
"loss": -0.4347,
|
| 802 |
+
"step": 360
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4604604604604605,
|
| 806 |
+
"grad_norm": 1.6063904762268066,
|
| 807 |
+
"grpo_mean_advantage": 4.313886279305734e-07,
|
| 808 |
+
"grpo_mean_group_score": 0.5884170532226562,
|
| 809 |
+
"grpo_mean_kl_div": 0.0,
|
| 810 |
+
"grpo_std_advantage": 1.9621948013082147e-06,
|
| 811 |
+
"learning_rate": 9.449718040529987e-07,
|
| 812 |
+
"loss": -0.6217,
|
| 813 |
+
"step": 365
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.4804804804804805,
|
| 817 |
+
"grad_norm": 2.114664077758789,
|
| 818 |
+
"grpo_mean_advantage": 2.0489096641540527e-07,
|
| 819 |
+
"grpo_mean_group_score": 0.5795440673828125,
|
| 820 |
+
"grpo_mean_kl_div": 0.0,
|
| 821 |
+
"grpo_std_advantage": 1.0235522722723545e-06,
|
| 822 |
+
"learning_rate": 8.810997742939531e-07,
|
| 823 |
+
"loss": -0.5364,
|
| 824 |
+
"step": 370
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.5005005005005005,
|
| 828 |
+
"grad_norm": 1.8450465202331543,
|
| 829 |
+
"grpo_mean_advantage": -1.4185905001795618e-06,
|
| 830 |
+
"grpo_mean_group_score": 0.5607603788375854,
|
| 831 |
+
"grpo_mean_kl_div": 0.0,
|
| 832 |
+
"grpo_std_advantage": 1.0947338523692451e-05,
|
| 833 |
+
"learning_rate": 8.189979881632634e-07,
|
| 834 |
+
"loss": -0.4798,
|
| 835 |
+
"step": 375
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.5205205205205206,
|
| 839 |
+
"grad_norm": 2.673438787460327,
|
| 840 |
+
"grpo_mean_advantage": -1.758337049295733e-07,
|
| 841 |
+
"grpo_mean_group_score": 0.5381432771682739,
|
| 842 |
+
"grpo_mean_kl_div": 0.0,
|
| 843 |
+
"grpo_std_advantage": 9.663675655247062e-07,
|
| 844 |
+
"learning_rate": 7.587343530522945e-07,
|
| 845 |
+
"loss": -0.4805,
|
| 846 |
+
"step": 380
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.5405405405405406,
|
| 850 |
+
"grad_norm": 2.2263550758361816,
|
| 851 |
+
"grpo_mean_advantage": -6.973743325033865e-07,
|
| 852 |
+
"grpo_mean_group_score": 0.5528443455696106,
|
| 853 |
+
"grpo_mean_kl_div": 0.0,
|
| 854 |
+
"grpo_std_advantage": 4.341973180999048e-06,
|
| 855 |
+
"learning_rate": 7.003747663612581e-07,
|
| 856 |
+
"loss": -0.433,
|
| 857 |
+
"step": 385
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 1.5605605605605606,
|
| 861 |
+
"grad_norm": 2.3657093048095703,
|
| 862 |
+
"grpo_mean_advantage": 1.7881394143159923e-08,
|
| 863 |
+
"grpo_mean_group_score": 0.6091476678848267,
|
| 864 |
+
"grpo_mean_kl_div": 0.0,
|
| 865 |
+
"grpo_std_advantage": 1.3004198251564958e-07,
|
| 866 |
+
"learning_rate": 6.439830434413754e-07,
|
| 867 |
+
"loss": -0.6021,
|
| 868 |
+
"step": 390
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.5805805805805806,
|
| 872 |
+
"grad_norm": 1.9847129583358765,
|
| 873 |
+
"grpo_mean_advantage": 3.4868716625169327e-07,
|
| 874 |
+
"grpo_mean_group_score": 0.5397372245788574,
|
| 875 |
+
"grpo_mean_kl_div": 0.0,
|
| 876 |
+
"grpo_std_advantage": 2.059372718576924e-06,
|
| 877 |
+
"learning_rate": 5.896208478137222e-07,
|
| 878 |
+
"loss": -0.5595,
|
| 879 |
+
"step": 395
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.6006006006006006,
|
| 883 |
+
"grad_norm": 2.922114133834839,
|
| 884 |
+
"grpo_mean_advantage": -2.1636485598719446e-06,
|
| 885 |
+
"grpo_mean_group_score": 0.5873125195503235,
|
| 886 |
+
"grpo_mean_kl_div": 0.0,
|
| 887 |
+
"grpo_std_advantage": 9.725940799398813e-06,
|
| 888 |
+
"learning_rate": 5.373476237410808e-07,
|
| 889 |
+
"loss": -0.5592,
|
| 890 |
+
"step": 400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.6206206206206206,
|
| 894 |
+
"grad_norm": 1.8524045944213867,
|
| 895 |
+
"grpo_mean_advantage": -5.960464477539063e-08,
|
| 896 |
+
"grpo_mean_group_score": 0.5601426362991333,
|
| 897 |
+
"grpo_mean_kl_div": 0.0,
|
| 898 |
+
"grpo_std_advantage": 3.460792754594877e-07,
|
| 899 |
+
"learning_rate": 4.872205312265074e-07,
|
| 900 |
+
"loss": -0.5623,
|
| 901 |
+
"step": 405
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.6406406406406406,
|
| 905 |
+
"grad_norm": 1.7269790172576904,
|
| 906 |
+
"grpo_mean_advantage": 2.6226044269606064e-07,
|
| 907 |
+
"grpo_mean_group_score": 0.578656792640686,
|
| 908 |
+
"grpo_mean_kl_div": 0.0,
|
| 909 |
+
"grpo_std_advantage": 7.928817922220333e-07,
|
| 910 |
+
"learning_rate": 4.3929438350970687e-07,
|
| 911 |
+
"loss": -0.5943,
|
| 912 |
+
"step": 410
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.6606606606606606,
|
| 916 |
+
"grad_norm": 2.26530122756958,
|
| 917 |
+
"grpo_mean_advantage": 2.3558736756967846e-06,
|
| 918 |
+
"grpo_mean_group_score": 0.5885810852050781,
|
| 919 |
+
"grpo_mean_kl_div": 0.0,
|
| 920 |
+
"grpo_std_advantage": 1.4469559573626611e-05,
|
| 921 |
+
"learning_rate": 3.936215871295634e-07,
|
| 922 |
+
"loss": -0.6193,
|
| 923 |
+
"step": 415
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.6806806806806807,
|
| 927 |
+
"grad_norm": 2.6794464588165283,
|
| 928 |
+
"grpo_mean_advantage": 1.639127766850379e-08,
|
| 929 |
+
"grpo_mean_group_score": 0.5805023312568665,
|
| 930 |
+
"grpo_mean_kl_div": 0.0,
|
| 931 |
+
"grpo_std_advantage": 9.352411325380672e-07,
|
| 932 |
+
"learning_rate": 3.502520846183577e-07,
|
| 933 |
+
"loss": -0.6934,
|
| 934 |
+
"step": 420
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7007007007007007,
|
| 938 |
+
"grad_norm": 2.100447654724121,
|
| 939 |
+
"grpo_mean_advantage": 3.2387674764322583e-06,
|
| 940 |
+
"grpo_mean_group_score": 0.5655918121337891,
|
| 941 |
+
"grpo_mean_kl_div": 0.0,
|
| 942 |
+
"grpo_std_advantage": 1.999079904635437e-05,
|
| 943 |
+
"learning_rate": 3.092332998903416e-07,
|
| 944 |
+
"loss": -0.5126,
|
| 945 |
+
"step": 425
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1.7207207207207207,
|
| 949 |
+
"grad_norm": 2.1027915477752686,
|
| 950 |
+
"grpo_mean_advantage": 3.5464762504489045e-07,
|
| 951 |
+
"grpo_mean_group_score": 0.5474504232406616,
|
| 952 |
+
"grpo_mean_kl_div": 0.0,
|
| 953 |
+
"grpo_std_advantage": 1.7663603557593888e-06,
|
| 954 |
+
"learning_rate": 2.706100863843822e-07,
|
| 955 |
+
"loss": -0.5446,
|
| 956 |
+
"step": 430
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.7407407407407407,
|
| 960 |
+
"grad_norm": 2.289045572280884,
|
| 961 |
+
"grpo_mean_advantage": 3.6135315895080566e-07,
|
| 962 |
+
"grpo_mean_group_score": 0.5874254703521729,
|
| 963 |
+
"grpo_mean_kl_div": 0.0,
|
| 964 |
+
"grpo_std_advantage": 2.356920958845876e-06,
|
| 965 |
+
"learning_rate": 2.3442467801738867e-07,
|
| 966 |
+
"loss": -0.5125,
|
| 967 |
+
"step": 435
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.7607607607607607,
|
| 971 |
+
"grad_norm": 2.278038501739502,
|
| 972 |
+
"grpo_mean_advantage": 2.7567148563889532e-08,
|
| 973 |
+
"grpo_mean_group_score": 0.5815118551254272,
|
| 974 |
+
"grpo_mean_kl_div": 0.0,
|
| 975 |
+
"grpo_std_advantage": 9.97340521280421e-07,
|
| 976 |
+
"learning_rate": 2.007166430021415e-07,
|
| 977 |
+
"loss": -0.595,
|
| 978 |
+
"step": 440
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.7807807807807807,
|
| 982 |
+
"grad_norm": 2.340942621231079,
|
| 983 |
+
"grpo_mean_advantage": -8.34465012644614e-08,
|
| 984 |
+
"grpo_mean_group_score": 0.5611211061477661,
|
| 985 |
+
"grpo_mean_kl_div": 0.0,
|
| 986 |
+
"grpo_std_advantage": 5.558832185670326e-07,
|
| 987 |
+
"learning_rate": 1.6952284058003366e-07,
|
| 988 |
+
"loss": -0.8055,
|
| 989 |
+
"step": 445
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.800800800800801,
|
| 993 |
+
"grad_norm": 2.4256298542022705,
|
| 994 |
+
"grpo_mean_advantage": -1.9818544672034477e-07,
|
| 995 |
+
"grpo_mean_group_score": 0.590424656867981,
|
| 996 |
+
"grpo_mean_kl_div": 0.0,
|
| 997 |
+
"grpo_std_advantage": 6.800727305744658e-07,
|
| 998 |
+
"learning_rate": 1.4087738071603075e-07,
|
| 999 |
+
"loss": -0.8561,
|
| 1000 |
+
"step": 450
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.820820820820821,
|
| 1004 |
+
"grad_norm": 1.6453255414962769,
|
| 1005 |
+
"grpo_mean_advantage": -1.9371508841459217e-08,
|
| 1006 |
+
"grpo_mean_group_score": 0.5670351982116699,
|
| 1007 |
+
"grpo_mean_kl_div": 0.0,
|
| 1008 |
+
"grpo_std_advantage": 3.142378943721269e-07,
|
| 1009 |
+
"learning_rate": 1.1481158679992554e-07,
|
| 1010 |
+
"loss": -0.429,
|
| 1011 |
+
"step": 455
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 1.840840840840841,
|
| 1015 |
+
"grad_norm": 2.3458049297332764,
|
| 1016 |
+
"grpo_mean_advantage": 2.3692845729783585e-07,
|
| 1017 |
+
"grpo_mean_group_score": 0.5640432834625244,
|
| 1018 |
+
"grpo_mean_kl_div": 0.0,
|
| 1019 |
+
"grpo_std_advantage": 1.682946731307311e-06,
|
| 1020 |
+
"learning_rate": 9.135396139467151e-08,
|
| 1021 |
+
"loss": -0.642,
|
| 1022 |
+
"step": 460
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.860860860860861,
|
| 1026 |
+
"grad_norm": 2.730945110321045,
|
| 1027 |
+
"grpo_mean_advantage": 1.110136480519941e-07,
|
| 1028 |
+
"grpo_mean_group_score": 0.5584251284599304,
|
| 1029 |
+
"grpo_mean_kl_div": 0.0,
|
| 1030 |
+
"grpo_std_advantage": 8.930008448260196e-07,
|
| 1031 |
+
"learning_rate": 7.053015506924749e-08,
|
| 1032 |
+
"loss": -0.583,
|
| 1033 |
+
"step": 465
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 1.880880880880881,
|
| 1037 |
+
"grad_norm": 2.1463465690612793,
|
| 1038 |
+
"grpo_mean_advantage": 2.5406478698641877e-07,
|
| 1039 |
+
"grpo_mean_group_score": 0.5392154455184937,
|
| 1040 |
+
"grpo_mean_kl_div": 0.0,
|
| 1041 |
+
"grpo_std_advantage": 9.93092498902115e-07,
|
| 1042 |
+
"learning_rate": 5.236293835013839e-08,
|
| 1043 |
+
"loss": -0.5197,
|
| 1044 |
+
"step": 470
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.900900900900901,
|
| 1048 |
+
"grad_norm": 2.427900791168213,
|
| 1049 |
+
"grpo_mean_advantage": -8.940697071579962e-09,
|
| 1050 |
+
"grpo_mean_group_score": 0.5686308741569519,
|
| 1051 |
+
"grpo_mean_kl_div": 0.0,
|
| 1052 |
+
"grpo_std_advantage": 5.835169645251881e-07,
|
| 1053 |
+
"learning_rate": 3.687217682209837e-08,
|
| 1054 |
+
"loss": -0.5864,
|
| 1055 |
+
"step": 475
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 1.920920920920921,
|
| 1059 |
+
"grad_norm": 2.042795419692993,
|
| 1060 |
+
"grpo_mean_advantage": 4.0605664253234863e-07,
|
| 1061 |
+
"grpo_mean_group_score": 0.5842767357826233,
|
| 1062 |
+
"grpo_mean_kl_div": 0.0,
|
| 1063 |
+
"grpo_std_advantage": 2.3210795916384086e-06,
|
| 1064 |
+
"learning_rate": 2.4074809405425227e-08,
|
| 1065 |
+
"loss": -0.5721,
|
| 1066 |
+
"step": 480
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.940940940940941,
|
| 1070 |
+
"grad_norm": 2.800136089324951,
|
| 1071 |
+
"grpo_mean_advantage": 1.341104507446289e-07,
|
| 1072 |
+
"grpo_mean_group_score": 0.5495311617851257,
|
| 1073 |
+
"grpo_mean_kl_div": 0.0,
|
| 1074 |
+
"grpo_std_advantage": 1.507950400991831e-06,
|
| 1075 |
+
"learning_rate": 1.3984829833499636e-08,
|
| 1076 |
+
"loss": -0.5944,
|
| 1077 |
+
"step": 485
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 1.960960960960961,
|
| 1081 |
+
"grad_norm": 2.8475866317749023,
|
| 1082 |
+
"grpo_mean_advantage": 2.689659481802664e-07,
|
| 1083 |
+
"grpo_mean_group_score": 0.549436628818512,
|
| 1084 |
+
"grpo_mean_kl_div": 0.0,
|
| 1085 |
+
"grpo_std_advantage": 8.491958851664094e-07,
|
| 1086 |
+
"learning_rate": 6.6132713508446075e-09,
|
| 1087 |
+
"loss": -0.7015,
|
| 1088 |
+
"step": 490
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.980980980980981,
|
| 1092 |
+
"grad_norm": 2.9422402381896973,
|
| 1093 |
+
"grpo_mean_advantage": 8.195638656616211e-08,
|
| 1094 |
+
"grpo_mean_group_score": 0.544632077217102,
|
| 1095 |
+
"grpo_mean_kl_div": 0.0,
|
| 1096 |
+
"grpo_std_advantage": 3.802849732892355e-06,
|
| 1097 |
+
"learning_rate": 1.9681946484320645e-09,
|
| 1098 |
+
"loss": -0.4033,
|
| 1099 |
+
"step": 495
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 2.0,
|
| 1103 |
+
"grad_norm": 2.66204833984375,
|
| 1104 |
+
"grpo_mean_advantage": 7.552536089860951e-07,
|
| 1105 |
+
"grpo_mean_group_score": 0.5968535542488098,
|
| 1106 |
+
"grpo_mean_kl_div": 0.0,
|
| 1107 |
+
"grpo_std_advantage": 4.143997102801222e-06,
|
| 1108 |
+
"learning_rate": 5.467904943851077e-11,
|
| 1109 |
+
"loss": -0.6773,
|
| 1110 |
+
"step": 500
|
| 1111 |
+
}
|
| 1112 |
+
],
|
| 1113 |
+
"logging_steps": 5,
|
| 1114 |
+
"max_steps": 500,
|
| 1115 |
+
"num_input_tokens_seen": 0,
|
| 1116 |
+
"num_train_epochs": 2,
|
| 1117 |
+
"save_steps": 100,
|
| 1118 |
+
"stateful_callbacks": {
|
| 1119 |
+
"TrainerControl": {
|
| 1120 |
+
"args": {
|
| 1121 |
+
"should_epoch_stop": false,
|
| 1122 |
+
"should_evaluate": false,
|
| 1123 |
+
"should_log": false,
|
| 1124 |
+
"should_save": true,
|
| 1125 |
+
"should_training_stop": true
|
| 1126 |
+
},
|
| 1127 |
+
"attributes": {}
|
| 1128 |
+
}
|
| 1129 |
+
},
|
| 1130 |
+
"total_flos": 0.0,
|
| 1131 |
+
"train_batch_size": 1,
|
| 1132 |
+
"trial_name": null,
|
| 1133 |
+
"trial_params": null
|
| 1134 |
+
}
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a4a4e48ed61b7c96f3bd2836ac828013a311834ab8a9542ea461fe1ff953396b
|
| 3 |
+
size 5496
|
grpo_qwen_14B_v2/checkpoints/checkpoint-500/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
grpo_qwen_14B_v2/config_resolved.yaml
ADDED
|
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
run:
|
| 2 |
+
run_dir: runs/grpo_qwen_14b
|
| 3 |
+
model:
|
| 4 |
+
repo_id: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 5 |
+
tokenizer_name: Qwen/Qwen2.5-Coder-14B
|
| 6 |
+
load_in_8bit: false
|
| 7 |
+
load_in_4bit: false
|
| 8 |
+
torch_dtype: bfloat16
|
| 9 |
+
device_map: auto
|
| 10 |
+
trust_remote_code: true
|
| 11 |
+
wandb:
|
| 12 |
+
enabled: true
|
| 13 |
+
project: rl-training
|
| 14 |
+
entity: null
|
| 15 |
+
name: null
|
| 16 |
+
tags:
|
| 17 |
+
- grpo-lora
|
| 18 |
+
- 14B-QWEN
|
| 19 |
+
notes: null
|
| 20 |
+
data:
|
| 21 |
+
train_jsonl: grpo_dataset.jsonl
|
| 22 |
+
eval_jsonl: null
|
| 23 |
+
eval_split_ratio: 0.0
|
| 24 |
+
shuffle: true
|
| 25 |
+
num_proc: 1
|
| 26 |
+
prompt_field: prompt
|
| 27 |
+
completions_field: completions
|
| 28 |
+
scores_field: scores
|
| 29 |
+
format_type: raw
|
| 30 |
+
max_length: 2048
|
| 31 |
+
min_completions: 2
|
| 32 |
+
system_prompt: "You are a Hyperswitch Rust code analyzer. Identify functions/structs\
|
| 33 |
+
\ that need modification for a given task.\n\n## Output Format\n\n##OUTPUT\nExplain\
|
| 34 |
+
\ the data flow and why each component must change:\n- Flow: [Input \u2192 Processing\
|
| 35 |
+
\ \u2192 Output with arrows]\n- For each component: \"The [ComponentName] ([path])\
|
| 36 |
+
\ must [action] because [reason]\u2014without this, [consequence]\"\n- Explain\
|
| 37 |
+
\ coupling between components\n\n##SELECT\nmodify::crates/path/to/file.rs::impl::ComponentName\n\
|
| 38 |
+
add::crates/another/file.rs::function::AnotherComponent\n<EOS>\n\n## Rules\n\n\
|
| 39 |
+
1. Use full paths: `remove::crates/folder/file.rs::Type::Name`\n2. Use `::` for\
|
| 40 |
+
\ nested items: `status::StructName::Type::Name`\n3. Always explain \"must change\
|
| 41 |
+
\ because\" and \"without this\"\n3. Types of components: function, struct, enum,\
|
| 42 |
+
\ impl, trait\n4. If there is extra information (e.g., enum variants), include\
|
| 43 |
+
\ that too.\n5. Start with ##OUTPUT, end with ##SELECT, terminate with <EOS>\n\
|
| 44 |
+
\n## Example\n\n##TASK\nAdd webhook subscription support\n\n##OUTPUT\nThe webhook\
|
| 45 |
+
\ system routes events via EventClass enum. Flow: webhook \u2192 EventClass \u2192\
|
| 46 |
+
\ handler \u2192 processing. The EventClass enum (crates/common_enums/src/enums.rs::EventClass)\
|
| 47 |
+
\ must add Subscriptions variant because it defines event routing\u2014without\
|
| 48 |
+
\ this, subscription events cannot be processed. The SubscriptionStatus impl (crates/common_enums/src/transformers.rs::SubscriptionStatus)\
|
| 49 |
+
\ must map to EventType because it converts status to events\u2014without this,\
|
| 50 |
+
\ status changes don't trigger webhooks. These are coupled: EventClass routes\
|
| 51 |
+
\ to handlers that use SubscriptionStatus mappings.\n\n##SELECT\ncrates/common_enums/src/enums.rs::EventClass\n\
|
| 52 |
+
crates/common_enums/src/transformers.rs::SubscriptionStatus\n<EOS>\n"
|
| 53 |
+
custom_template: '##INSTRUCTION
|
| 54 |
+
|
| 55 |
+
{instruction}<|im_end|>
|
| 56 |
+
|
| 57 |
+
{input}<|im_end|>
|
| 58 |
+
|
| 59 |
+
{output}<|im_end|>'
|
| 60 |
+
grpo:
|
| 61 |
+
group_size: 4
|
| 62 |
+
kl_coef: 0.05
|
| 63 |
+
normalize_advantages: true
|
| 64 |
+
reward_scaling: 1.0
|
| 65 |
+
reward_bias: 0.0
|
| 66 |
+
reward_clip: 5.0
|
| 67 |
+
advantage_temperature: 1.0
|
| 68 |
+
use_reference_model: false
|
| 69 |
+
seed: 42
|
| 70 |
+
peft:
|
| 71 |
+
enabled: true
|
| 72 |
+
r: 16
|
| 73 |
+
lora_alpha: 32
|
| 74 |
+
lora_dropout: 0.05
|
| 75 |
+
target_modules:
|
| 76 |
+
- q_proj
|
| 77 |
+
- k_proj
|
| 78 |
+
- v_proj
|
| 79 |
+
- o_proj
|
| 80 |
+
- gate_proj
|
| 81 |
+
- up_proj
|
| 82 |
+
- down_proj
|
| 83 |
+
bias: none
|
| 84 |
+
task_type: CAUSAL_LM
|
| 85 |
+
train:
|
| 86 |
+
output_dir: runs/grpo_14b_run1
|
| 87 |
+
num_train_epochs: 2
|
| 88 |
+
per_device_train_batch_size: 1
|
| 89 |
+
gradient_accumulation_steps: 8
|
| 90 |
+
per_device_eval_batch_size: 1
|
| 91 |
+
learning_rate: 5.0e-06
|
| 92 |
+
weight_decay: 0.01
|
| 93 |
+
warmup_ratio: 0.05
|
| 94 |
+
lr_scheduler_type: cosine
|
| 95 |
+
fp16: false
|
| 96 |
+
bf16: true
|
| 97 |
+
max_grad_norm: 1.0
|
| 98 |
+
gradient_checkpointing: true
|
| 99 |
+
logging_steps: 5
|
| 100 |
+
save_steps: 100
|
| 101 |
+
save_total_limit: 2
|
| 102 |
+
evaluation_strategy: 'no'
|
| 103 |
+
dataloader_num_workers: 4
|
| 104 |
+
dataloader_pin_memory: true
|
| 105 |
+
remove_unused_columns: false
|
| 106 |
+
report_to: []
|
| 107 |
+
seed: 42
|
| 108 |
+
ddp_find_unused_parameters: false
|
| 109 |
+
merge:
|
| 110 |
+
enabled: true
|
| 111 |
+
merged_dtype: float16
|
| 112 |
+
max_shard_size: 2GB
|
| 113 |
+
output_dir: ./Models/Qwen-Coder-14B-HS-CPT-SFT-v2-GRPO
|
| 114 |
+
upload:
|
| 115 |
+
enabled: false
|
grpo_qwen_14B_v2/logs/grpo_metrics.jsonl
ADDED
|
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"ts": "2025-12-27T19:40:44", "step": 5, "epoch": 0.02, "grpo_mean_advantage": -1.3560057254835556e-07, "grpo_std_advantage": 3.0318567496578908e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5922331809997559}
|
| 2 |
+
{"ts": "2025-12-27T19:46:55", "step": 5, "epoch": 0.02, "grpo_mean_advantage": -1.3560057254835556e-07, "grpo_std_advantage": 3.0318567496578908e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5922331809997559}
|
| 3 |
+
{"ts": "2025-12-27T19:48:18", "step": 10, "epoch": 0.04, "grpo_mean_advantage": 3.6619603633880615e-06, "grpo_std_advantage": 1.6246918676188216e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5561589002609253}
|
| 4 |
+
{"ts": "2025-12-27T19:49:40", "step": 15, "epoch": 0.0601, "grpo_mean_advantage": -1.0654330395709621e-07, "grpo_std_advantage": 5.399440965447866e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5759152173995972}
|
| 5 |
+
{"ts": "2025-12-27T19:51:01", "step": 20, "epoch": 0.0801, "grpo_mean_advantage": -5.871057737749652e-07, "grpo_std_advantage": 2.6951597646984737e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5127314329147339}
|
| 6 |
+
{"ts": "2025-12-27T19:52:23", "step": 25, "epoch": 0.1001, "grpo_mean_advantage": 6.370246410369873e-07, "grpo_std_advantage": 2.8908377771585947e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.539706826210022}
|
| 7 |
+
{"ts": "2025-12-27T19:53:48", "step": 30, "epoch": 0.1201, "grpo_mean_advantage": 6.705522359595761e-09, "grpo_std_advantage": 6.189450800775376e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5812538862228394}
|
| 8 |
+
{"ts": "2025-12-27T19:55:13", "step": 35, "epoch": 0.1401, "grpo_mean_advantage": 3.859400692363124e-07, "grpo_std_advantage": 1.6833292875162442e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5909844636917114}
|
| 9 |
+
{"ts": "2025-12-27T19:56:38", "step": 40, "epoch": 0.1602, "grpo_mean_advantage": 2.600252742013254e-07, "grpo_std_advantage": 1.4095899132371414e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5630953907966614}
|
| 10 |
+
{"ts": "2025-12-27T19:58:00", "step": 45, "epoch": 0.1802, "grpo_mean_advantage": -1.2591480924584175e-07, "grpo_std_advantage": 1.0309080380466185e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5604403614997864}
|
| 11 |
+
{"ts": "2025-12-27T19:59:27", "step": 50, "epoch": 0.2002, "grpo_mean_advantage": -2.808868941883702e-07, "grpo_std_advantage": 1.5696078889959608e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5971035957336426}
|
| 12 |
+
{"ts": "2025-12-27T20:00:53", "step": 55, "epoch": 0.2202, "grpo_mean_advantage": 2.6822089438383045e-08, "grpo_std_advantage": 3.7878271541558206e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5892971754074097}
|
| 13 |
+
{"ts": "2025-12-27T20:02:17", "step": 60, "epoch": 0.2402, "grpo_mean_advantage": -5.662441182607836e-08, "grpo_std_advantage": 6.128998393251095e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.564322292804718}
|
| 14 |
+
{"ts": "2025-12-27T20:03:45", "step": 65, "epoch": 0.2603, "grpo_mean_advantage": -1.5944242193199898e-07, "grpo_std_advantage": 1.6374274309782777e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.562497615814209}
|
| 15 |
+
{"ts": "2025-12-27T20:05:10", "step": 70, "epoch": 0.2803, "grpo_mean_advantage": 1.6838312433264946e-07, "grpo_std_advantage": 8.536571272088622e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5904761552810669}
|
| 16 |
+
{"ts": "2025-12-27T20:06:36", "step": 75, "epoch": 0.3003, "grpo_mean_advantage": 1.1175870895385742e-07, "grpo_std_advantage": 6.451961667153228e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5765624046325684}
|
| 17 |
+
{"ts": "2025-12-27T20:08:03", "step": 80, "epoch": 0.3203, "grpo_mean_advantage": -1.4603138254187797e-07, "grpo_std_advantage": 1.1309343790344428e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5858271718025208}
|
| 18 |
+
{"ts": "2025-12-27T20:09:31", "step": 85, "epoch": 0.3403, "grpo_mean_advantage": -1.817941665649414e-06, "grpo_std_advantage": 1.1141768482048064e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5871662497520447}
|
| 19 |
+
{"ts": "2025-12-27T20:10:55", "step": 90, "epoch": 0.3604, "grpo_mean_advantage": 1.8179416372277046e-07, "grpo_std_advantage": 6.210335072864837e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5330992937088013}
|
| 20 |
+
{"ts": "2025-12-27T20:12:19", "step": 95, "epoch": 0.3804, "grpo_mean_advantage": -2.972781771859445e-07, "grpo_std_advantage": 3.1582342217006953e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5265295505523682}
|
| 21 |
+
{"ts": "2025-12-27T20:13:46", "step": 100, "epoch": 0.4004, "grpo_mean_advantage": -7.033348197182931e-07, "grpo_std_advantage": 4.245831405569334e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5660771131515503}
|
| 22 |
+
{"ts": "2025-12-27T20:15:15", "step": 105, "epoch": 0.4204, "grpo_mean_advantage": 1.1920928955078125e-07, "grpo_std_advantage": 3.2809634831210133e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.57631915807724}
|
| 23 |
+
{"ts": "2025-12-27T20:16:39", "step": 110, "epoch": 0.4404, "grpo_mean_advantage": -4.0978193283081055e-07, "grpo_std_advantage": 6.0397578636184335e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.546563982963562}
|
| 24 |
+
{"ts": "2025-12-27T20:18:01", "step": 115, "epoch": 0.4605, "grpo_mean_advantage": -1.467764434437413e-07, "grpo_std_advantage": 2.2689375782647403e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5519219636917114}
|
| 25 |
+
{"ts": "2025-12-27T20:19:25", "step": 120, "epoch": 0.4805, "grpo_mean_advantage": -5.215406329028838e-09, "grpo_std_advantage": 7.929010621410271e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5490407943725586}
|
| 26 |
+
{"ts": "2025-12-27T20:20:48", "step": 125, "epoch": 0.5005, "grpo_mean_advantage": -5.7369469175228005e-08, "grpo_std_advantage": 1.2823379620385822e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5646580457687378}
|
| 27 |
+
{"ts": "2025-12-27T20:22:09", "step": 130, "epoch": 0.5205, "grpo_mean_advantage": 2.9876827056796174e-07, "grpo_std_advantage": 1.0496698905626545e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6111599802970886}
|
| 28 |
+
{"ts": "2025-12-27T20:23:32", "step": 135, "epoch": 0.5405, "grpo_mean_advantage": 1.5869736103013565e-07, "grpo_std_advantage": 1.2748531617035042e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5619662404060364}
|
| 29 |
+
{"ts": "2025-12-27T20:24:57", "step": 140, "epoch": 0.5606, "grpo_mean_advantage": 3.0100346748440643e-07, "grpo_std_advantage": 2.4499684059264837e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5795454978942871}
|
| 30 |
+
{"ts": "2025-12-27T20:26:19", "step": 145, "epoch": 0.5806, "grpo_mean_advantage": -3.233552092751779e-07, "grpo_std_advantage": 1.248456669600273e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5804953575134277}
|
| 31 |
+
{"ts": "2025-12-27T20:27:43", "step": 150, "epoch": 0.6006, "grpo_mean_advantage": 3.2261013416245987e-07, "grpo_std_advantage": 1.4773489738217904e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5628539323806763}
|
| 32 |
+
{"ts": "2025-12-27T20:29:07", "step": 155, "epoch": 0.6206, "grpo_mean_advantage": -2.5331974029541016e-07, "grpo_std_advantage": 1.5092309695319273e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5727725625038147}
|
| 33 |
+
{"ts": "2025-12-27T20:30:30", "step": 160, "epoch": 0.6406, "grpo_mean_advantage": -6.780028627417778e-08, "grpo_std_advantage": 8.550978805033083e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5833909511566162}
|
| 34 |
+
{"ts": "2025-12-27T20:31:51", "step": 165, "epoch": 0.6607, "grpo_mean_advantage": -5.587935447692871e-08, "grpo_std_advantage": 3.564579174053506e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5742615461349487}
|
| 35 |
+
{"ts": "2025-12-27T20:33:17", "step": 170, "epoch": 0.6807, "grpo_mean_advantage": -5.327165126800537e-07, "grpo_std_advantage": 2.309018327650847e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5758188962936401}
|
| 36 |
+
{"ts": "2025-12-27T20:34:40", "step": 175, "epoch": 0.7007, "grpo_mean_advantage": 5.863606702405377e-07, "grpo_std_advantage": 2.4449204829579685e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5767683982849121}
|
| 37 |
+
{"ts": "2025-12-27T20:36:09", "step": 180, "epoch": 0.7207, "grpo_mean_advantage": 3.2186508747145126e-07, "grpo_std_advantage": 2.293551688126172e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.586772084236145}
|
| 38 |
+
{"ts": "2025-12-27T20:37:33", "step": 185, "epoch": 0.7407, "grpo_mean_advantage": -4.470348358154297e-08, "grpo_std_advantage": 3.7067667335577426e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.549396276473999}
|
| 39 |
+
{"ts": "2025-12-27T20:38:56", "step": 190, "epoch": 0.7608, "grpo_mean_advantage": -2.1010637851759384e-07, "grpo_std_advantage": 1.1695076409523608e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5798425078392029}
|
| 40 |
+
{"ts": "2025-12-27T20:40:21", "step": 195, "epoch": 0.7808, "grpo_mean_advantage": 1.765787658314366e-07, "grpo_std_advantage": 2.429934738756856e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5584167838096619}
|
| 41 |
+
{"ts": "2025-12-27T20:41:46", "step": 200, "epoch": 0.8008, "grpo_mean_advantage": 1.6540289493605087e-07, "grpo_std_advantage": 2.6342788714828203e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5676193237304688}
|
| 42 |
+
{"ts": "2025-12-27T20:43:13", "step": 205, "epoch": 0.8208, "grpo_mean_advantage": -1.0944902442133753e-06, "grpo_std_advantage": 5.346942998585291e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5669739842414856}
|
| 43 |
+
{"ts": "2025-12-27T20:44:38", "step": 210, "epoch": 0.8408, "grpo_mean_advantage": 2.4065374759629776e-07, "grpo_std_advantage": 1.6327536513927043e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5922158360481262}
|
| 44 |
+
{"ts": "2025-12-27T20:46:02", "step": 215, "epoch": 0.8609, "grpo_mean_advantage": -5.21540641784668e-08, "grpo_std_advantage": 5.847922466273303e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5473950505256653}
|
| 45 |
+
{"ts": "2025-12-27T20:47:28", "step": 220, "epoch": 0.8809, "grpo_mean_advantage": 6.541609991472797e-07, "grpo_std_advantage": 4.072162937518442e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5880032777786255}
|
| 46 |
+
{"ts": "2025-12-27T20:48:53", "step": 225, "epoch": 0.9009, "grpo_mean_advantage": -1.2218951894737984e-07, "grpo_std_advantage": 4.386006935419573e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5835092663764954}
|
| 47 |
+
{"ts": "2025-12-27T20:50:20", "step": 230, "epoch": 0.9209, "grpo_mean_advantage": 1.7605722177904681e-06, "grpo_std_advantage": 8.007580618141219e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5394966006278992}
|
| 48 |
+
{"ts": "2025-12-27T20:51:41", "step": 235, "epoch": 0.9409, "grpo_mean_advantage": -3.3080578987210174e-07, "grpo_std_advantage": 1.551636614749441e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5687432289123535}
|
| 49 |
+
{"ts": "2025-12-27T20:53:06", "step": 240, "epoch": 0.961, "grpo_mean_advantage": 2.712011450967111e-07, "grpo_std_advantage": 1.4400844747797237e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5550583600997925}
|
| 50 |
+
{"ts": "2025-12-27T20:54:27", "step": 245, "epoch": 0.981, "grpo_mean_advantage": -3.2857059295565705e-07, "grpo_std_advantage": 2.105091425619321e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.558111310005188}
|
| 51 |
+
{"ts": "2025-12-27T20:55:45", "step": 250, "epoch": 1.0, "grpo_mean_advantage": 4.470348358154297e-08, "grpo_std_advantage": 5.315724820320611e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6196198463439941}
|
| 52 |
+
{"ts": "2025-12-27T20:57:06", "step": 255, "epoch": 1.02, "grpo_mean_advantage": 9.290873776990338e-07, "grpo_std_advantage": 4.219644324621186e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.582168459892273}
|
| 53 |
+
{"ts": "2025-12-27T20:58:26", "step": 260, "epoch": 1.04, "grpo_mean_advantage": 2.533197474008375e-08, "grpo_std_advantage": 1.6600588992332632e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5551307797431946}
|
| 54 |
+
{"ts": "2025-12-27T20:59:48", "step": 265, "epoch": 1.0601, "grpo_mean_advantage": -5.662441182607836e-08, "grpo_std_advantage": 1.0909400316450046e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.535040020942688}
|
| 55 |
+
{"ts": "2025-12-27T21:01:15", "step": 270, "epoch": 1.0801, "grpo_mean_advantage": -9.536743306171047e-08, "grpo_std_advantage": 5.838213610331877e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5673571825027466}
|
| 56 |
+
{"ts": "2025-12-27T21:02:37", "step": 275, "epoch": 1.1001, "grpo_mean_advantage": 3.278255533700758e-08, "grpo_std_advantage": 9.317170679423725e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5874732732772827}
|
| 57 |
+
{"ts": "2025-12-27T21:03:59", "step": 280, "epoch": 1.1201, "grpo_mean_advantage": -1.206994113545079e-07, "grpo_std_advantage": 6.201085511747806e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5569106340408325}
|
| 58 |
+
{"ts": "2025-12-27T21:05:19", "step": 285, "epoch": 1.1401, "grpo_mean_advantage": 4.470348358154297e-08, "grpo_std_advantage": 6.115651558502577e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5578873157501221}
|
| 59 |
+
{"ts": "2025-12-27T21:06:43", "step": 290, "epoch": 1.1602, "grpo_mean_advantage": -3.3453108017056365e-07, "grpo_std_advantage": 3.5326345368957845e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5735999345779419}
|
| 60 |
+
{"ts": "2025-12-27T21:08:10", "step": 295, "epoch": 1.1802, "grpo_mean_advantage": -1.110136480519941e-07, "grpo_std_advantage": 4.731904823529476e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5626259446144104}
|
| 61 |
+
{"ts": "2025-12-27T21:09:34", "step": 300, "epoch": 1.2002, "grpo_mean_advantage": -5.08874677507265e-07, "grpo_std_advantage": 1.840126174101897e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5463050603866577}
|
| 62 |
+
{"ts": "2025-12-27T21:10:59", "step": 305, "epoch": 1.2202, "grpo_mean_advantage": 1.01327898960335e-07, "grpo_std_advantage": 7.798533943059738e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5352144241333008}
|
| 63 |
+
{"ts": "2025-12-27T21:12:25", "step": 310, "epoch": 1.2402, "grpo_mean_advantage": 1.341104507446289e-07, "grpo_std_advantage": 7.821902840987605e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5547868013381958}
|
| 64 |
+
{"ts": "2025-12-27T21:13:52", "step": 315, "epoch": 1.2603, "grpo_mean_advantage": 9.015202806494926e-08, "grpo_std_advantage": 1.0693488547985908e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5859472751617432}
|
| 65 |
+
{"ts": "2025-12-27T21:15:21", "step": 320, "epoch": 1.2803, "grpo_mean_advantage": -2.443790378947597e-07, "grpo_std_advantage": 1.183122208203713e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5751550793647766}
|
| 66 |
+
{"ts": "2025-12-27T21:16:47", "step": 325, "epoch": 1.3003, "grpo_mean_advantage": -6.705522537231445e-08, "grpo_std_advantage": 6.109748937888071e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5497723817825317}
|
| 67 |
+
{"ts": "2025-12-27T21:18:13", "step": 330, "epoch": 1.3203, "grpo_mean_advantage": -1.639127766850379e-08, "grpo_std_advantage": 5.529495297196263e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.55989670753479}
|
| 68 |
+
{"ts": "2025-12-27T21:19:37", "step": 335, "epoch": 1.3403, "grpo_mean_advantage": 4.418194237132411e-07, "grpo_std_advantage": 2.9275292945385445e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5809233784675598}
|
| 69 |
+
{"ts": "2025-12-27T21:20:59", "step": 340, "epoch": 1.3604, "grpo_mean_advantage": 9.685754776000977e-08, "grpo_std_advantage": 3.754235251562932e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5568087100982666}
|
| 70 |
+
{"ts": "2025-12-27T21:22:27", "step": 345, "epoch": 1.3804, "grpo_mean_advantage": -2.384185791015625e-07, "grpo_std_advantage": 6.821086913078034e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5655568838119507}
|
| 71 |
+
{"ts": "2025-12-27T21:23:49", "step": 350, "epoch": 1.4004, "grpo_mean_advantage": -8.717179156292332e-08, "grpo_std_advantage": 2.500940354366321e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6089578866958618}
|
| 72 |
+
{"ts": "2025-12-27T21:25:13", "step": 355, "epoch": 1.4204, "grpo_mean_advantage": 2.1606683731079102e-07, "grpo_std_advantage": 1.4568390724889468e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6129671335220337}
|
| 73 |
+
{"ts": "2025-12-27T21:26:36", "step": 360, "epoch": 1.4404, "grpo_mean_advantage": -3.725290298461914e-09, "grpo_std_advantage": 2.965894054796081e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5562310814857483}
|
| 74 |
+
{"ts": "2025-12-27T21:27:58", "step": 365, "epoch": 1.4605, "grpo_mean_advantage": 4.313886279305734e-07, "grpo_std_advantage": 1.9621948013082147e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5884170532226562}
|
| 75 |
+
{"ts": "2025-12-27T21:29:21", "step": 370, "epoch": 1.4805, "grpo_mean_advantage": 2.0489096641540527e-07, "grpo_std_advantage": 1.0235522722723545e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5795440673828125}
|
| 76 |
+
{"ts": "2025-12-27T21:30:45", "step": 375, "epoch": 1.5005, "grpo_mean_advantage": -1.4185905001795618e-06, "grpo_std_advantage": 1.0947338523692451e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5607603788375854}
|
| 77 |
+
{"ts": "2025-12-27T21:32:13", "step": 380, "epoch": 1.5205, "grpo_mean_advantage": -1.758337049295733e-07, "grpo_std_advantage": 9.663675655247062e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5381432771682739}
|
| 78 |
+
{"ts": "2025-12-27T21:33:36", "step": 385, "epoch": 1.5405, "grpo_mean_advantage": -6.973743325033865e-07, "grpo_std_advantage": 4.341973180999048e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5528443455696106}
|
| 79 |
+
{"ts": "2025-12-27T21:34:59", "step": 390, "epoch": 1.5606, "grpo_mean_advantage": 1.7881394143159923e-08, "grpo_std_advantage": 1.3004198251564958e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6091476678848267}
|
| 80 |
+
{"ts": "2025-12-27T21:36:21", "step": 395, "epoch": 1.5806, "grpo_mean_advantage": 3.4868716625169327e-07, "grpo_std_advantage": 2.059372718576924e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5397372245788574}
|
| 81 |
+
{"ts": "2025-12-27T21:37:48", "step": 400, "epoch": 1.6006, "grpo_mean_advantage": -2.1636485598719446e-06, "grpo_std_advantage": 9.725940799398813e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5873125195503235}
|
| 82 |
+
{"ts": "2025-12-27T21:39:11", "step": 405, "epoch": 1.6206, "grpo_mean_advantage": -5.960464477539063e-08, "grpo_std_advantage": 3.460792754594877e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5601426362991333}
|
| 83 |
+
{"ts": "2025-12-27T21:40:38", "step": 410, "epoch": 1.6406, "grpo_mean_advantage": 2.6226044269606064e-07, "grpo_std_advantage": 7.928817922220333e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.578656792640686}
|
| 84 |
+
{"ts": "2025-12-27T21:42:02", "step": 415, "epoch": 1.6607, "grpo_mean_advantage": 2.3558736756967846e-06, "grpo_std_advantage": 1.4469559573626611e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5885810852050781}
|
| 85 |
+
{"ts": "2025-12-27T21:43:27", "step": 420, "epoch": 1.6807, "grpo_mean_advantage": 1.639127766850379e-08, "grpo_std_advantage": 9.352411325380672e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5805023312568665}
|
| 86 |
+
{"ts": "2025-12-27T21:44:52", "step": 425, "epoch": 1.7007, "grpo_mean_advantage": 3.2387674764322583e-06, "grpo_std_advantage": 1.999079904635437e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5655918121337891}
|
| 87 |
+
{"ts": "2025-12-27T21:46:17", "step": 430, "epoch": 1.7207, "grpo_mean_advantage": 3.5464762504489045e-07, "grpo_std_advantage": 1.7663603557593888e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5474504232406616}
|
| 88 |
+
{"ts": "2025-12-27T21:47:43", "step": 435, "epoch": 1.7407, "grpo_mean_advantage": 3.6135315895080566e-07, "grpo_std_advantage": 2.356920958845876e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5874254703521729}
|
| 89 |
+
{"ts": "2025-12-27T21:49:08", "step": 440, "epoch": 1.7608, "grpo_mean_advantage": 2.7567148563889532e-08, "grpo_std_advantage": 9.97340521280421e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5815118551254272}
|
| 90 |
+
{"ts": "2025-12-27T21:50:35", "step": 445, "epoch": 1.7808, "grpo_mean_advantage": -8.34465012644614e-08, "grpo_std_advantage": 5.558832185670326e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5611211061477661}
|
| 91 |
+
{"ts": "2025-12-27T21:52:06", "step": 450, "epoch": 1.8008, "grpo_mean_advantage": -1.9818544672034477e-07, "grpo_std_advantage": 6.800727305744658e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.590424656867981}
|
| 92 |
+
{"ts": "2025-12-27T21:53:32", "step": 455, "epoch": 1.8208, "grpo_mean_advantage": -1.9371508841459217e-08, "grpo_std_advantage": 3.142378943721269e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5670351982116699}
|
| 93 |
+
{"ts": "2025-12-27T21:55:00", "step": 460, "epoch": 1.8408, "grpo_mean_advantage": 2.3692845729783585e-07, "grpo_std_advantage": 1.682946731307311e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5640432834625244}
|
| 94 |
+
{"ts": "2025-12-27T21:56:27", "step": 465, "epoch": 1.8609, "grpo_mean_advantage": 1.110136480519941e-07, "grpo_std_advantage": 8.930008448260196e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5584251284599304}
|
| 95 |
+
{"ts": "2025-12-27T21:57:50", "step": 470, "epoch": 1.8809, "grpo_mean_advantage": 2.5406478698641877e-07, "grpo_std_advantage": 9.93092498902115e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5392154455184937}
|
| 96 |
+
{"ts": "2025-12-27T21:59:15", "step": 475, "epoch": 1.9009, "grpo_mean_advantage": -8.940697071579962e-09, "grpo_std_advantage": 5.835169645251881e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5686308741569519}
|
| 97 |
+
{"ts": "2025-12-27T22:00:36", "step": 480, "epoch": 1.9209, "grpo_mean_advantage": 4.0605664253234863e-07, "grpo_std_advantage": 2.3210795916384086e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5842767357826233}
|
| 98 |
+
{"ts": "2025-12-27T22:01:58", "step": 485, "epoch": 1.9409, "grpo_mean_advantage": 1.341104507446289e-07, "grpo_std_advantage": 1.507950400991831e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5495311617851257}
|
| 99 |
+
{"ts": "2025-12-27T22:03:20", "step": 490, "epoch": 1.961, "grpo_mean_advantage": 2.689659481802664e-07, "grpo_std_advantage": 8.491958851664094e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.549436628818512}
|
| 100 |
+
{"ts": "2025-12-27T22:04:42", "step": 495, "epoch": 1.981, "grpo_mean_advantage": 8.195638656616211e-08, "grpo_std_advantage": 3.802849732892355e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.544632077217102}
|
| 101 |
+
{"ts": "2025-12-27T22:06:01", "step": 500, "epoch": 2.0, "grpo_mean_advantage": 7.552536089860951e-07, "grpo_std_advantage": 4.143997102801222e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5968535542488098}
|
grpo_qwen_14B_v2/logs/train.jsonl
ADDED
|
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"ts": "2025-12-27T19:40:44", "event": "train_log", "step": 5, "epoch": 0.02002002002002002, "progress_pct": 1.0, "epoch_pct": 1.0, "eta": "02:23:23", "max_grad_norm": 1.0, "loss": 0.007, "grad_norm": 0.06052486598491669, "learning_rate": 8.000000000000001e-07, "grpo_mean_advantage": -1.3560057254835556e-07, "grpo_std_advantage": 3.0318567496578908e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5922331809997559}
|
| 2 |
+
{"ts": "2025-12-27T19:46:55", "event": "train_log", "step": 5, "epoch": 0.02002002002002002, "progress_pct": 1.0, "epoch_pct": 1.0, "eta": "02:16:53", "max_grad_norm": 1.0, "loss": 0.007, "grad_norm": 0.05460292845964432, "learning_rate": 8.000000000000001e-07, "grpo_mean_advantage": -1.3560057254835556e-07, "grpo_std_advantage": 3.0318567496578908e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5922331809997559}
|
| 3 |
+
{"ts": "2025-12-27T19:48:18", "event": "train_log", "step": 10, "epoch": 0.04004004004004004, "progress_pct": 2.0, "epoch_pct": 2.0, "eta": "02:15:24", "max_grad_norm": 1.0, "loss": 0.0107, "grad_norm": 0.0679207444190979, "learning_rate": 1.8000000000000001e-06, "grpo_mean_advantage": 3.6619603633880615e-06, "grpo_std_advantage": 1.6246918676188216e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5561589002609253}
|
| 4 |
+
{"ts": "2025-12-27T19:49:40", "event": "train_log", "step": 15, "epoch": 0.06006006006006006, "progress_pct": 3.0, "epoch_pct": 3.0, "eta": "02:13:25", "max_grad_norm": 1.0, "loss": 0.007, "grad_norm": 0.05788416787981987, "learning_rate": 2.8000000000000003e-06, "grpo_mean_advantage": -1.0654330395709621e-07, "grpo_std_advantage": 5.399440965447866e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5759152173995972}
|
| 5 |
+
{"ts": "2025-12-27T19:51:01", "event": "train_log", "step": 20, "epoch": 0.08008008008008008, "progress_pct": 4.0, "epoch_pct": 4.0, "eta": "02:11:33", "max_grad_norm": 1.0, "loss": 0.0246, "grad_norm": 0.0746568813920021, "learning_rate": 3.8000000000000005e-06, "grpo_mean_advantage": -5.871057737749652e-07, "grpo_std_advantage": 2.6951597646984737e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5127314329147339}
|
| 6 |
+
{"ts": "2025-12-27T19:52:23", "event": "train_log", "step": 25, "epoch": 0.1001001001001001, "progress_pct": 5.0, "epoch_pct": 5.01, "eta": "02:10:02", "max_grad_norm": 1.0, "loss": 0.0337, "grad_norm": 0.11442846059799194, "learning_rate": 4.800000000000001e-06, "grpo_mean_advantage": 6.370246410369873e-07, "grpo_std_advantage": 2.8908377771585947e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.539706826210022}
|
| 7 |
+
{"ts": "2025-12-27T19:53:48", "event": "train_log", "step": 30, "epoch": 0.12012012012012012, "progress_pct": 6.0, "epoch_pct": 6.01, "eta": "02:09:24", "max_grad_norm": 1.0, "loss": 0.0171, "grad_norm": 0.05778791010379791, "learning_rate": 4.999125183044924e-06, "grpo_mean_advantage": 6.705522359595761e-09, "grpo_std_advantage": 6.189450800775376e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5812538862228394}
|
| 8 |
+
{"ts": "2025-12-27T19:55:13", "event": "train_log", "step": 35, "epoch": 0.14014014014014015, "progress_pct": 7.0, "epoch_pct": 7.01, "eta": "02:08:36", "max_grad_norm": 1.0, "loss": 0.0145, "grad_norm": 0.05819695070385933, "learning_rate": 4.995572288443412e-06, "grpo_mean_advantage": 3.859400692363124e-07, "grpo_std_advantage": 1.6833292875162442e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5909844636917114}
|
| 9 |
+
{"ts": "2025-12-27T19:56:38", "event": "train_log", "step": 40, "epoch": 0.16016016016016016, "progress_pct": 8.0, "epoch_pct": 8.01, "eta": "02:07:30", "max_grad_norm": 1.0, "loss": 0.0196, "grad_norm": 0.07968433201313019, "learning_rate": 4.98929052218411e-06, "grpo_mean_advantage": 2.600252742013254e-07, "grpo_std_advantage": 1.4095899132371414e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5630953907966614}
|
| 10 |
+
{"ts": "2025-12-27T19:58:00", "event": "train_log", "step": 45, "epoch": 0.18018018018018017, "progress_pct": 9.0, "epoch_pct": 9.01, "eta": "02:06:04", "max_grad_norm": 1.0, "loss": 0.0186, "grad_norm": 0.0733402892947197, "learning_rate": 4.980286753286196e-06, "grpo_mean_advantage": -1.2591480924584175e-07, "grpo_std_advantage": 1.0309080380466185e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5604403614997864}
|
| 11 |
+
{"ts": "2025-12-27T19:59:27", "event": "train_log", "step": 50, "epoch": 0.2002002002002002, "progress_pct": 10.0, "epoch_pct": 10.01, "eta": "02:05:12", "max_grad_norm": 1.0, "loss": 0.0286, "grad_norm": 0.07136482000350952, "learning_rate": 4.9685708272387645e-06, "grpo_mean_advantage": -2.808868941883702e-07, "grpo_std_advantage": 1.5696078889959608e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5971035957336426}
|
| 12 |
+
{"ts": "2025-12-27T20:00:53", "event": "train_log", "step": 55, "epoch": 0.22022022022022023, "progress_pct": 11.0, "epoch_pct": 11.01, "eta": "02:04:07", "max_grad_norm": 1.0, "loss": 0.0054, "grad_norm": 0.08851475268602371, "learning_rate": 4.9541555552349404e-06, "grpo_mean_advantage": 2.6822089438383045e-08, "grpo_std_advantage": 3.7878271541558206e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5892971754074097}
|
| 13 |
+
{"ts": "2025-12-27T20:02:17", "event": "train_log", "step": 60, "epoch": 0.24024024024024024, "progress_pct": 12.0, "epoch_pct": 12.01, "eta": "02:02:51", "max_grad_norm": 1.0, "loss": -0.0074, "grad_norm": 0.07778509706258774, "learning_rate": 4.9370567001630155e-06, "grpo_mean_advantage": -5.662441182607836e-08, "grpo_std_advantage": 6.128998393251095e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.564322292804718}
|
| 14 |
+
{"ts": "2025-12-27T20:03:45", "event": "train_log", "step": 65, "epoch": 0.2602602602602603, "progress_pct": 13.0, "epoch_pct": 13.01, "eta": "02:01:55", "max_grad_norm": 1.0, "loss": 0.0145, "grad_norm": 0.08740051090717316, "learning_rate": 4.917292959369968e-06, "grpo_mean_advantage": -1.5944242193199898e-07, "grpo_std_advantage": 1.6374274309782777e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.562497615814209}
|
| 15 |
+
{"ts": "2025-12-27T20:05:10", "event": "train_log", "step": 70, "epoch": 0.2802802802802803, "progress_pct": 14.0, "epoch_pct": 14.01, "eta": "02:00:32", "max_grad_norm": 1.0, "loss": 0.0257, "grad_norm": 0.19070060551166534, "learning_rate": 4.8948859442161876e-06, "grpo_mean_advantage": 1.6838312433264946e-07, "grpo_std_advantage": 8.536571272088622e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5904761552810669}
|
| 16 |
+
{"ts": "2025-12-27T20:06:36", "event": "train_log", "step": 75, "epoch": 0.3003003003003003, "progress_pct": 15.0, "epoch_pct": 15.02, "eta": "01:59:23", "max_grad_norm": 1.0, "loss": 0.0024, "grad_norm": 0.07321271300315857, "learning_rate": 4.869860156443768e-06, "grpo_mean_advantage": 1.1175870895385742e-07, "grpo_std_advantage": 6.451961667153228e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5765624046325684}
|
| 17 |
+
{"ts": "2025-12-27T20:08:03", "event": "train_log", "step": 80, "epoch": 0.3203203203203203, "progress_pct": 16.0, "epoch_pct": 16.02, "eta": "01:58:12", "max_grad_norm": 1.0, "loss": 0.0277, "grad_norm": 0.07126748561859131, "learning_rate": 4.842242961384211e-06, "grpo_mean_advantage": -1.4603138254187797e-07, "grpo_std_advantage": 1.1309343790344428e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5858271718025208}
|
| 18 |
+
{"ts": "2025-12-27T20:09:31", "event": "train_log", "step": 85, "epoch": 0.34034034034034033, "progress_pct": 17.0, "epoch_pct": 17.02, "eta": "01:57:04", "max_grad_norm": 1.0, "loss": 0.0246, "grad_norm": 0.08629189431667328, "learning_rate": 4.812064558034847e-06, "grpo_mean_advantage": -1.817941665649414e-06, "grpo_std_advantage": 1.1141768482048064e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5871662497520447}
|
| 19 |
+
{"ts": "2025-12-27T20:10:55", "event": "train_log", "step": 90, "epoch": 0.36036036036036034, "progress_pct": 18.0, "epoch_pct": 18.02, "eta": "01:55:38", "max_grad_norm": 1.0, "loss": 0.0056, "grad_norm": 0.0998779758810997, "learning_rate": 4.779357946036662e-06, "grpo_mean_advantage": 1.8179416372277046e-07, "grpo_std_advantage": 6.210335072864837e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5330992937088013}
|
| 20 |
+
{"ts": "2025-12-27T20:12:19", "event": "train_log", "step": 95, "epoch": 0.38038038038038036, "progress_pct": 19.0, "epoch_pct": 19.02, "eta": "01:54:09", "max_grad_norm": 1.0, "loss": 0.0053, "grad_norm": 0.10614689439535141, "learning_rate": 4.74415888958968e-06, "grpo_mean_advantage": -2.972781771859445e-07, "grpo_std_advantage": 3.1582342217006953e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5265295505523682}
|
| 21 |
+
{"ts": "2025-12-27T20:13:46", "event": "train_log", "step": 100, "epoch": 0.4004004004004004, "progress_pct": 20.0, "epoch_pct": 20.02, "eta": "01:52:55", "max_grad_norm": 1.0, "loss": 0.0134, "grad_norm": 0.10345634073019028, "learning_rate": 4.706505878345343e-06, "grpo_mean_advantage": -7.033348197182931e-07, "grpo_std_advantage": 4.245831405569334e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5660771131515503}
|
| 22 |
+
{"ts": "2025-12-27T20:15:15", "event": "train_log", "step": 105, "epoch": 0.42042042042042044, "progress_pct": 21.0, "epoch_pct": 21.02, "eta": "01:51:45", "max_grad_norm": 1.0, "loss": 0.0004, "grad_norm": 0.10077933222055435, "learning_rate": 4.666440085318626e-06, "grpo_mean_advantage": 1.1920928955078125e-07, "grpo_std_advantage": 3.2809634831210133e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.57631915807724}
|
| 23 |
+
{"ts": "2025-12-27T20:16:39", "event": "train_log", "step": 110, "epoch": 0.44044044044044045, "progress_pct": 22.0, "epoch_pct": 22.02, "eta": "01:50:19", "max_grad_norm": 1.0, "loss": 0.0033, "grad_norm": 0.09548182785511017, "learning_rate": 4.624005321865968e-06, "grpo_mean_advantage": -4.0978193283081055e-07, "grpo_std_advantage": 6.0397578636184335e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.546563982963562}
|
| 24 |
+
{"ts": "2025-12-27T20:18:01", "event": "train_log", "step": 115, "epoch": 0.46046046046046046, "progress_pct": 23.0, "epoch_pct": 23.02, "eta": "01:48:43", "max_grad_norm": 1.0, "loss": 0.0095, "grad_norm": 0.09417816251516342, "learning_rate": 4.57924798977818e-06, "grpo_mean_advantage": -1.467764434437413e-07, "grpo_std_advantage": 2.2689375782647403e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5519219636917114}
|
| 25 |
+
{"ts": "2025-12-27T20:19:25", "event": "train_log", "step": 120, "epoch": 0.4804804804804805, "progress_pct": 24.0, "epoch_pct": 24.02, "eta": "01:47:18", "max_grad_norm": 1.0, "loss": 0.0006, "grad_norm": 0.10022275149822235, "learning_rate": 4.532217030540781e-06, "grpo_mean_advantage": -5.215406329028838e-09, "grpo_std_advantage": 7.929010621410271e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5490407943725586}
|
| 26 |
+
{"ts": "2025-12-27T20:20:48", "event": "train_log", "step": 125, "epoch": 0.5005005005005005, "progress_pct": 25.0, "epoch_pct": 25.03, "eta": "01:45:46", "max_grad_norm": 1.0, "loss": -0.0046, "grad_norm": 0.14057794213294983, "learning_rate": 4.482963871817195e-06, "grpo_mean_advantage": -5.7369469175228005e-08, "grpo_std_advantage": 1.2823379620385822e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5646580457687378}
|
| 27 |
+
{"ts": "2025-12-27T20:22:09", "event": "train_log", "step": 130, "epoch": 0.5205205205205206, "progress_pct": 26.0, "epoch_pct": 26.03, "eta": "01:44:12", "max_grad_norm": 1.0, "loss": -0.003, "grad_norm": 0.12420658767223358, "learning_rate": 4.4315423712133595e-06, "grpo_mean_advantage": 2.9876827056796174e-07, "grpo_std_advantage": 1.0496698905626545e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6111599802970886}
|
| 28 |
+
{"ts": "2025-12-27T20:23:32", "event": "train_log", "step": 135, "epoch": 0.5405405405405406, "progress_pct": 27.0, "epoch_pct": 27.03, "eta": "01:42:43", "max_grad_norm": 1.0, "loss": 0.0154, "grad_norm": 0.14342808723449707, "learning_rate": 4.378008757385222e-06, "grpo_mean_advantage": 1.5869736103013565e-07, "grpo_std_advantage": 1.2748531617035042e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5619662404060364}
|
| 29 |
+
{"ts": "2025-12-27T20:24:57", "event": "train_log", "step": 140, "epoch": 0.5605605605605606, "progress_pct": 28.0, "epoch_pct": 28.03, "eta": "01:41:21", "max_grad_norm": 1.0, "loss": -0.0262, "grad_norm": 0.14729444682598114, "learning_rate": 4.322421568553529e-06, "grpo_mean_advantage": 3.0100346748440643e-07, "grpo_std_advantage": 2.4499684059264837e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5795454978942871}
|
| 30 |
+
{"ts": "2025-12-27T20:26:19", "event": "train_log", "step": 145, "epoch": 0.5805805805805806, "progress_pct": 29.0, "epoch_pct": 29.03, "eta": "01:39:51", "max_grad_norm": 1.0, "loss": 0.0018, "grad_norm": 0.15249410271644592, "learning_rate": 4.2648415884931476e-06, "grpo_mean_advantage": -3.233552092751779e-07, "grpo_std_advantage": 1.248456669600273e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5804953575134277}
|
| 31 |
+
{"ts": "2025-12-27T20:27:43", "event": "train_log", "step": 150, "epoch": 0.6006006006006006, "progress_pct": 30.0, "epoch_pct": 30.03, "eta": "01:38:25", "max_grad_norm": 1.0, "loss": -0.017, "grad_norm": 0.1841023564338684, "learning_rate": 4.205331780066892e-06, "grpo_mean_advantage": 3.2261013416245987e-07, "grpo_std_advantage": 1.4773489738217904e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5628539323806763}
|
| 32 |
+
{"ts": "2025-12-27T20:29:07", "event": "train_log", "step": 155, "epoch": 0.6206206206206206, "progress_pct": 31.0, "epoch_pct": 31.03, "eta": "01:37:00", "max_grad_norm": 1.0, "loss": 0.0044, "grad_norm": 0.18597163259983063, "learning_rate": 4.1439572163765615e-06, "grpo_mean_advantage": -2.5331974029541016e-07, "grpo_std_advantage": 1.5092309695319273e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5727725625038147}
|
| 33 |
+
{"ts": "2025-12-27T20:30:30", "event": "train_log", "step": 160, "epoch": 0.6406406406406406, "progress_pct": 32.0, "epoch_pct": 32.03, "eta": "01:35:33", "max_grad_norm": 1.0, "loss": -0.005, "grad_norm": 0.18310388922691345, "learning_rate": 4.0807850096064605e-06, "grpo_mean_advantage": -6.780028627417778e-08, "grpo_std_advantage": 8.550978805033083e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5833909511566162}
|
| 34 |
+
{"ts": "2025-12-27T20:31:51", "event": "train_log", "step": 165, "epoch": 0.6606606606606606, "progress_pct": 33.0, "epoch_pct": 33.03, "eta": "01:34:01", "max_grad_norm": 1.0, "loss": -0.015, "grad_norm": 0.2192923128604889, "learning_rate": 4.015884237637206e-06, "grpo_mean_advantage": -5.587935447692871e-08, "grpo_std_advantage": 3.564579174053506e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5742615461349487}
|
| 35 |
+
{"ts": "2025-12-27T20:33:17", "event": "train_log", "step": 170, "epoch": 0.6806806806806807, "progress_pct": 34.0, "epoch_pct": 34.03, "eta": "01:32:41", "max_grad_norm": 1.0, "loss": -0.0314, "grad_norm": 0.16708803176879883, "learning_rate": 3.949325868510083e-06, "grpo_mean_advantage": -5.327165126800537e-07, "grpo_std_advantage": 2.309018327650847e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5758188962936401}
|
| 36 |
+
{"ts": "2025-12-27T20:34:40", "event": "train_log", "step": 175, "epoch": 0.7007007007007007, "progress_pct": 35.0, "epoch_pct": 35.04, "eta": "01:31:14", "max_grad_norm": 1.0, "loss": -0.0441, "grad_norm": 0.3401262164115906, "learning_rate": 3.881182682824534e-06, "grpo_mean_advantage": 5.863606702405377e-07, "grpo_std_advantage": 2.4449204829579685e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5767683982849121}
|
| 37 |
+
{"ts": "2025-12-27T20:36:09", "event": "train_log", "step": 180, "epoch": 0.7207207207207207, "progress_pct": 36.0, "epoch_pct": 36.04, "eta": "01:29:58", "max_grad_norm": 1.0, "loss": -0.0162, "grad_norm": 0.1931898146867752, "learning_rate": 3.811529194153635e-06, "grpo_mean_advantage": 3.2186508747145126e-07, "grpo_std_advantage": 2.293551688126172e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.586772084236145}
|
| 38 |
+
{"ts": "2025-12-27T20:37:33", "event": "train_log", "step": 185, "epoch": 0.7407407407407407, "progress_pct": 37.0, "epoch_pct": 37.04, "eta": "01:28:33", "max_grad_norm": 1.0, "loss": -0.0386, "grad_norm": 0.2537969648838043, "learning_rate": 3.7404415675646054e-06, "grpo_mean_advantage": -4.470348358154297e-08, "grpo_std_advantage": 3.7067667335577426e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.549396276473999}
|
| 39 |
+
{"ts": "2025-12-27T20:38:56", "event": "train_log", "step": 190, "epoch": 0.7607607607607607, "progress_pct": 38.0, "epoch_pct": 38.04, "eta": "01:27:06", "max_grad_norm": 1.0, "loss": -0.037, "grad_norm": 0.20326584577560425, "learning_rate": 3.667997536333424e-06, "grpo_mean_advantage": -2.1010637851759384e-07, "grpo_std_advantage": 1.1695076409523608e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5798425078392029}
|
| 40 |
+
{"ts": "2025-12-27T20:40:21", "event": "train_log", "step": 195, "epoch": 0.7807807807807807, "progress_pct": 39.0, "epoch_pct": 39.04, "eta": "01:25:43", "max_grad_norm": 1.0, "loss": -0.0292, "grad_norm": 0.25048357248306274, "learning_rate": 3.59427631694463e-06, "grpo_mean_advantage": 1.765787658314366e-07, "grpo_std_advantage": 2.429934738756856e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5584167838096619}
|
| 41 |
+
{"ts": "2025-12-27T20:41:46", "event": "train_log", "step": 200, "epoch": 0.8008008008008008, "progress_pct": 40.0, "epoch_pct": 40.04, "eta": "01:24:20", "max_grad_norm": 1.0, "loss": -0.0454, "grad_norm": 0.2687569260597229, "learning_rate": 3.5193585224692595e-06, "grpo_mean_advantage": 1.6540289493605087e-07, "grpo_std_advantage": 2.6342788714828203e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5676193237304688}
|
| 42 |
+
{"ts": "2025-12-27T20:43:13", "event": "train_log", "step": 205, "epoch": 0.8208208208208209, "progress_pct": 41.0, "epoch_pct": 41.04, "eta": "01:22:59", "max_grad_norm": 1.0, "loss": -0.0423, "grad_norm": 0.22301620244979858, "learning_rate": 3.44332607441564e-06, "grpo_mean_advantage": -1.0944902442133753e-06, "grpo_std_advantage": 5.346942998585291e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5669739842414856}
|
| 43 |
+
{"ts": "2025-12-27T20:44:38", "event": "train_log", "step": 210, "epoch": 0.8408408408408409, "progress_pct": 42.0, "epoch_pct": 42.04, "eta": "01:21:37", "max_grad_norm": 1.0, "loss": -0.0857, "grad_norm": 0.3040211498737335, "learning_rate": 3.3662621131494204e-06, "grpo_mean_advantage": 2.4065374759629776e-07, "grpo_std_advantage": 1.6327536513927043e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5922158360481262}
|
| 44 |
+
{"ts": "2025-12-27T20:46:02", "event": "train_log", "step": 215, "epoch": 0.8608608608608609, "progress_pct": 43.0, "epoch_pct": 43.04, "eta": "01:20:11", "max_grad_norm": 1.0, "loss": -0.0278, "grad_norm": 0.27231141924858093, "learning_rate": 3.2882509069808044e-06, "grpo_mean_advantage": -5.21540641784668e-08, "grpo_std_advantage": 5.847922466273303e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5473950505256653}
|
| 45 |
+
{"ts": "2025-12-27T20:47:28", "event": "train_log", "step": 220, "epoch": 0.8808808808808809, "progress_pct": 44.0, "epoch_pct": 44.04, "eta": "01:18:49", "max_grad_norm": 1.0, "loss": -0.0727, "grad_norm": 0.3571636378765106, "learning_rate": 3.2093777600183873e-06, "grpo_mean_advantage": 6.541609991472797e-07, "grpo_std_advantage": 4.072162937518442e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5880032777786255}
|
| 46 |
+
{"ts": "2025-12-27T20:48:53", "event": "train_log", "step": 225, "epoch": 0.9009009009009009, "progress_pct": 45.0, "epoch_pct": 45.05, "eta": "01:17:25", "max_grad_norm": 1.0, "loss": -0.0464, "grad_norm": 0.306273490190506, "learning_rate": 3.1297289188903705e-06, "grpo_mean_advantage": -1.2218951894737984e-07, "grpo_std_advantage": 4.386006935419573e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5835092663764954}
|
| 47 |
+
{"ts": "2025-12-27T20:50:20", "event": "train_log", "step": 230, "epoch": 0.9209209209209209, "progress_pct": 46.0, "epoch_pct": 46.05, "eta": "01:16:03", "max_grad_norm": 1.0, "loss": -0.0295, "grad_norm": 0.2700377106666565, "learning_rate": 3.049391478435133e-06, "grpo_mean_advantage": 1.7605722177904681e-06, "grpo_std_advantage": 8.007580618141219e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5394966006278992}
|
| 48 |
+
{"ts": "2025-12-27T20:51:41", "event": "train_log", "step": 235, "epoch": 0.9409409409409409, "progress_pct": 47.0, "epoch_pct": 47.05, "eta": "01:14:35", "max_grad_norm": 1.0, "loss": -0.031, "grad_norm": 0.39531761407852173, "learning_rate": 2.9684532864643123e-06, "grpo_mean_advantage": -3.3080578987210174e-07, "grpo_std_advantage": 1.551636614749441e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5687432289123535}
|
| 49 |
+
{"ts": "2025-12-27T20:53:06", "event": "train_log", "step": 240, "epoch": 0.960960960960961, "progress_pct": 48.0, "epoch_pct": 48.05, "eta": "01:13:11", "max_grad_norm": 1.0, "loss": -0.0789, "grad_norm": 0.5987040996551514, "learning_rate": 2.887002847702504e-06, "grpo_mean_advantage": 2.712011450967111e-07, "grpo_std_advantage": 1.4400844747797237e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5550583600997925}
|
| 50 |
+
{"ts": "2025-12-27T20:54:27", "event": "train_log", "step": 245, "epoch": 0.980980980980981, "progress_pct": 49.0, "epoch_pct": 49.05, "eta": "01:11:42", "max_grad_norm": 1.0, "loss": -0.1131, "grad_norm": 0.5680716037750244, "learning_rate": 2.8051292270086506e-06, "grpo_mean_advantage": -3.2857059295565705e-07, "grpo_std_advantage": 2.105091425619321e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.558111310005188}
|
| 51 |
+
{"ts": "2025-12-27T20:55:45", "event": "train_log", "step": 250, "epoch": 1.0, "progress_pct": 50.0, "epoch_pct": 50.0, "eta": "01:10:13", "max_grad_norm": 1.0, "loss": -0.2232, "grad_norm": 0.6204046010971069, "learning_rate": 2.722921951984927e-06, "grpo_mean_advantage": 4.470348358154297e-08, "grpo_std_advantage": 5.315724820320611e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6196198463439941}
|
| 52 |
+
{"ts": "2025-12-27T20:57:06", "event": "train_log", "step": 255, "epoch": 1.02002002002002, "progress_pct": 51.0, "epoch_pct": 51.0, "eta": "01:08:45", "max_grad_norm": 1.0, "loss": -0.1363, "grad_norm": 0.8389026522636414, "learning_rate": 2.640470915079614e-06, "grpo_mean_advantage": 9.290873776990338e-07, "grpo_std_advantage": 4.219644324621186e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.582168459892273}
|
| 53 |
+
{"ts": "2025-12-27T20:58:26", "event": "train_log", "step": 260, "epoch": 1.04004004004004, "progress_pct": 52.0, "epoch_pct": 52.0, "eta": "01:07:17", "max_grad_norm": 1.0, "loss": -0.1868, "grad_norm": 0.9067686796188354, "learning_rate": 2.557866275291035e-06, "grpo_mean_advantage": 2.533197474008375e-08, "grpo_std_advantage": 1.6600588992332632e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5551307797431946}
|
| 54 |
+
{"ts": "2025-12-27T20:59:48", "event": "train_log", "step": 265, "epoch": 1.06006006006006, "progress_pct": 53.0, "epoch_pct": 53.0, "eta": "01:05:51", "max_grad_norm": 1.0, "loss": -0.1792, "grad_norm": 0.9277902841567993, "learning_rate": 2.4751983595800093e-06, "grpo_mean_advantage": -5.662441182607836e-08, "grpo_std_advantage": 1.0909400316450046e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.535040020942688}
|
| 55 |
+
{"ts": "2025-12-27T21:01:15", "event": "train_log", "step": 270, "epoch": 1.08008008008008, "progress_pct": 54.0, "epoch_pct": 54.0, "eta": "01:04:29", "max_grad_norm": 1.0, "loss": -0.1691, "grad_norm": 1.0715463161468506, "learning_rate": 2.392557564098649e-06, "grpo_mean_advantage": -9.536743306171047e-08, "grpo_std_advantage": 5.838213610331877e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5673571825027466}
|
| 56 |
+
{"ts": "2025-12-27T21:02:37", "event": "train_log", "step": 275, "epoch": 1.1001001001001, "progress_pct": 55.0, "epoch_pct": 55.01, "eta": "01:03:03", "max_grad_norm": 1.0, "loss": -0.1655, "grad_norm": 0.7759184837341309, "learning_rate": 2.3100342553434924e-06, "grpo_mean_advantage": 3.278255533700758e-08, "grpo_std_advantage": 9.317170679423725e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5874732732772827}
|
| 57 |
+
{"ts": "2025-12-27T21:03:59", "event": "train_log", "step": 280, "epoch": 1.12012012012012, "progress_pct": 56.0, "epoch_pct": 56.01, "eta": "01:01:38", "max_grad_norm": 1.0, "loss": -0.1821, "grad_norm": 0.9387398958206177, "learning_rate": 2.2277186713410688e-06, "grpo_mean_advantage": -1.206994113545079e-07, "grpo_std_advantage": 6.201085511747806e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5569106340408325}
|
| 58 |
+
{"ts": "2025-12-27T21:05:19", "event": "train_log", "step": 285, "epoch": 1.14014014014014, "progress_pct": 57.0, "epoch_pct": 57.01, "eta": "01:00:10", "max_grad_norm": 1.0, "loss": -0.2102, "grad_norm": 1.6132302284240723, "learning_rate": 2.1457008229739395e-06, "grpo_mean_advantage": 4.470348358154297e-08, "grpo_std_advantage": 6.115651558502577e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5578873157501221}
|
| 59 |
+
{"ts": "2025-12-27T21:06:43", "event": "train_log", "step": 290, "epoch": 1.16016016016016, "progress_pct": 58.0, "epoch_pct": 58.01, "eta": "00:58:47", "max_grad_norm": 1.0, "loss": -0.2937, "grad_norm": 0.8679026961326599, "learning_rate": 2.0640703955551214e-06, "grpo_mean_advantage": -3.3453108017056365e-07, "grpo_std_advantage": 3.5326345368957845e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5735999345779419}
|
| 60 |
+
{"ts": "2025-12-27T21:08:10", "event": "train_log", "step": 295, "epoch": 1.1801801801801801, "progress_pct": 59.0, "epoch_pct": 59.01, "eta": "00:57:24", "max_grad_norm": 1.0, "loss": -0.2598, "grad_norm": 1.0550166368484497, "learning_rate": 1.9829166507585084e-06, "grpo_mean_advantage": -1.110136480519941e-07, "grpo_std_advantage": 4.731904823529476e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5626259446144104}
|
| 61 |
+
{"ts": "2025-12-27T21:09:34", "event": "train_log", "step": 300, "epoch": 1.2002002002002001, "progress_pct": 60.0, "epoch_pct": 60.01, "eta": "00:56:00", "max_grad_norm": 1.0, "loss": -0.2546, "grad_norm": 1.2819372415542603, "learning_rate": 1.90232832901255e-06, "grpo_mean_advantage": -5.08874677507265e-07, "grpo_std_advantage": 1.840126174101897e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5463050603866577}
|
| 62 |
+
{"ts": "2025-12-27T21:10:59", "event": "train_log", "step": 305, "epoch": 1.2202202202202201, "progress_pct": 61.0, "epoch_pct": 61.01, "eta": "00:54:37", "max_grad_norm": 1.0, "loss": -0.1809, "grad_norm": 1.0188143253326416, "learning_rate": 1.82239355246389e-06, "grpo_mean_advantage": 1.01327898960335e-07, "grpo_std_advantage": 7.798533943059738e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5352144241333008}
|
| 63 |
+
{"ts": "2025-12-27T21:12:25", "event": "train_log", "step": 310, "epoch": 1.2402402402402402, "progress_pct": 62.0, "epoch_pct": 62.01, "eta": "00:53:14", "max_grad_norm": 1.0, "loss": -0.3559, "grad_norm": 2.0709052085876465, "learning_rate": 1.7431997286170923e-06, "grpo_mean_advantage": 1.341104507446289e-07, "grpo_std_advantage": 7.821902840987605e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5547868013381958}
|
| 64 |
+
{"ts": "2025-12-27T21:13:52", "event": "train_log", "step": 315, "epoch": 1.2602602602602602, "progress_pct": 63.0, "epoch_pct": 63.01, "eta": "00:51:52", "max_grad_norm": 1.0, "loss": -0.3874, "grad_norm": 1.8516215085983276, "learning_rate": 1.6648334547558227e-06, "grpo_mean_advantage": 9.015202806494926e-08, "grpo_std_advantage": 1.0693488547985908e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5859472751617432}
|
| 65 |
+
{"ts": "2025-12-27T21:15:21", "event": "train_log", "step": 320, "epoch": 1.2802802802802802, "progress_pct": 64.0, "epoch_pct": 64.01, "eta": "00:50:31", "max_grad_norm": 1.0, "loss": -0.3467, "grad_norm": 1.283104419708252, "learning_rate": 1.5873804232499862e-06, "grpo_mean_advantage": -2.443790378947597e-07, "grpo_std_advantage": 1.183122208203713e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5751550793647766}
|
| 66 |
+
{"ts": "2025-12-27T21:16:47", "event": "train_log", "step": 325, "epoch": 1.3003003003003002, "progress_pct": 65.0, "epoch_pct": 65.02, "eta": "00:49:07", "max_grad_norm": 1.0, "loss": -0.1703, "grad_norm": 1.4108576774597168, "learning_rate": 1.51092532785238e-06, "grpo_mean_advantage": -6.705522537231445e-08, "grpo_std_advantage": 6.109748937888071e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5497723817825317}
|
| 67 |
+
{"ts": "2025-12-27T21:18:13", "event": "train_log", "step": 330, "epoch": 1.3203203203203202, "progress_pct": 66.0, "epoch_pct": 66.02, "eta": "00:47:44", "max_grad_norm": 1.0, "loss": -0.2918, "grad_norm": 1.0421361923217773, "learning_rate": 1.4355517710873184e-06, "grpo_mean_advantage": -1.639127766850379e-08, "grpo_std_advantage": 5.529495297196263e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.55989670753479}
|
| 68 |
+
{"ts": "2025-12-27T21:19:37", "event": "train_log", "step": 335, "epoch": 1.3403403403403402, "progress_pct": 67.0, "epoch_pct": 67.02, "eta": "00:46:20", "max_grad_norm": 1.0, "loss": -0.3069, "grad_norm": 1.3465828895568848, "learning_rate": 1.361342172832502e-06, "grpo_mean_advantage": 4.418194237132411e-07, "grpo_std_advantage": 2.9275292945385445e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5809233784675598}
|
| 69 |
+
{"ts": "2025-12-27T21:20:59", "event": "train_log", "step": 340, "epoch": 1.3603603603603602, "progress_pct": 68.0, "epoch_pct": 68.02, "eta": "00:44:55", "max_grad_norm": 1.0, "loss": -0.5594, "grad_norm": 1.1959459781646729, "learning_rate": 1.2883776801940884e-06, "grpo_mean_advantage": 9.685754776000977e-08, "grpo_std_advantage": 3.754235251562932e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5568087100982666}
|
| 70 |
+
{"ts": "2025-12-27T21:22:27", "event": "train_log", "step": 345, "epoch": 1.3803803803803802, "progress_pct": 69.0, "epoch_pct": 69.02, "eta": "00:43:32", "max_grad_norm": 1.0, "loss": -0.4102, "grad_norm": 1.8967422246932983, "learning_rate": 1.216738078773522e-06, "grpo_mean_advantage": -2.384185791015625e-07, "grpo_std_advantage": 6.821086913078034e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5655568838119507}
|
| 71 |
+
{"ts": "2025-12-27T21:23:49", "event": "train_log", "step": 350, "epoch": 1.4004004004004005, "progress_pct": 70.0, "epoch_pct": 70.02, "eta": "00:42:06", "max_grad_norm": 1.0, "loss": -0.338, "grad_norm": 2.221132755279541, "learning_rate": 1.146501705423155e-06, "grpo_mean_advantage": -8.717179156292332e-08, "grpo_std_advantage": 2.500940354366321e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6089578866958618}
|
| 72 |
+
{"ts": "2025-12-27T21:25:13", "event": "train_log", "step": 355, "epoch": 1.4204204204204205, "progress_pct": 71.0, "epoch_pct": 71.02, "eta": "00:40:42", "max_grad_norm": 1.0, "loss": -0.4985, "grad_norm": 2.3640377521514893, "learning_rate": 1.0777453625860474e-06, "grpo_mean_advantage": 2.1606683731079102e-07, "grpo_std_advantage": 1.4568390724889468e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6129671335220337}
|
| 73 |
+
{"ts": "2025-12-27T21:26:36", "event": "train_log", "step": 360, "epoch": 1.4404404404404405, "progress_pct": 72.0, "epoch_pct": 72.02, "eta": "00:39:18", "max_grad_norm": 1.0, "loss": -0.4347, "grad_norm": 1.9084734916687012, "learning_rate": 1.0105442343136184e-06, "grpo_mean_advantage": -3.725290298461914e-09, "grpo_std_advantage": 2.965894054796081e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5562310814857483}
|
| 74 |
+
{"ts": "2025-12-27T21:27:58", "event": "train_log", "step": 365, "epoch": 1.4604604604604605, "progress_pct": 73.0, "epoch_pct": 73.02, "eta": "00:37:52", "max_grad_norm": 1.0, "loss": -0.6217, "grad_norm": 1.6063904762268066, "learning_rate": 9.449718040529987e-07, "grpo_mean_advantage": 4.313886279305734e-07, "grpo_std_advantage": 1.9621948013082147e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5884170532226562}
|
| 75 |
+
{"ts": "2025-12-27T21:29:21", "event": "train_log", "step": 370, "epoch": 1.4804804804804805, "progress_pct": 74.0, "epoch_pct": 74.02, "eta": "00:36:28", "max_grad_norm": 1.0, "loss": -0.5364, "grad_norm": 2.114664077758789, "learning_rate": 8.810997742939531e-07, "grpo_mean_advantage": 2.0489096641540527e-07, "grpo_std_advantage": 1.0235522722723545e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5795440673828125}
|
| 76 |
+
{"ts": "2025-12-27T21:30:45", "event": "train_log", "step": 375, "epoch": 1.5005005005005005, "progress_pct": 75.0, "epoch_pct": 75.03, "eta": "00:35:04", "max_grad_norm": 1.0, "loss": -0.4798, "grad_norm": 1.8450465202331543, "learning_rate": 8.189979881632634e-07, "grpo_mean_advantage": -1.4185905001795618e-06, "grpo_std_advantage": 1.0947338523692451e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5607603788375854}
|
| 77 |
+
{"ts": "2025-12-27T21:32:13", "event": "train_log", "step": 380, "epoch": 1.5205205205205206, "progress_pct": 76.0, "epoch_pct": 76.03, "eta": "00:33:41", "max_grad_norm": 1.0, "loss": -0.4805, "grad_norm": 2.673438787460327, "learning_rate": 7.587343530522945e-07, "grpo_mean_advantage": -1.758337049295733e-07, "grpo_std_advantage": 9.663675655247062e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5381432771682739}
|
| 78 |
+
{"ts": "2025-12-27T21:33:36", "event": "train_log", "step": 385, "epoch": 1.5405405405405406, "progress_pct": 77.0, "epoch_pct": 77.03, "eta": "00:32:16", "max_grad_norm": 1.0, "loss": -0.433, "grad_norm": 2.2263550758361816, "learning_rate": 7.003747663612581e-07, "grpo_mean_advantage": -6.973743325033865e-07, "grpo_std_advantage": 4.341973180999048e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5528443455696106}
|
| 79 |
+
{"ts": "2025-12-27T21:34:59", "event": "train_log", "step": 390, "epoch": 1.5605605605605606, "progress_pct": 78.0, "epoch_pct": 78.03, "eta": "00:30:52", "max_grad_norm": 1.0, "loss": -0.6021, "grad_norm": 2.3657093048095703, "learning_rate": 6.439830434413754e-07, "grpo_mean_advantage": 1.7881394143159923e-08, "grpo_std_advantage": 1.3004198251564958e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.6091476678848267}
|
| 80 |
+
{"ts": "2025-12-27T21:36:21", "event": "train_log", "step": 395, "epoch": 1.5805805805805806, "progress_pct": 79.0, "epoch_pct": 79.03, "eta": "00:29:27", "max_grad_norm": 1.0, "loss": -0.5595, "grad_norm": 1.9847129583358765, "learning_rate": 5.896208478137222e-07, "grpo_mean_advantage": 3.4868716625169327e-07, "grpo_std_advantage": 2.059372718576924e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5397372245788574}
|
| 81 |
+
{"ts": "2025-12-27T21:37:48", "event": "train_log", "step": 400, "epoch": 1.6006006006006006, "progress_pct": 80.0, "epoch_pct": 80.03, "eta": "00:28:03", "max_grad_norm": 1.0, "loss": -0.5592, "grad_norm": 2.922114133834839, "learning_rate": 5.373476237410808e-07, "grpo_mean_advantage": -2.1636485598719446e-06, "grpo_std_advantage": 9.725940799398813e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5873125195503235}
|
| 82 |
+
{"ts": "2025-12-27T21:39:11", "event": "train_log", "step": 405, "epoch": 1.6206206206206206, "progress_pct": 81.0, "epoch_pct": 81.03, "eta": "00:26:39", "max_grad_norm": 1.0, "loss": -0.5623, "grad_norm": 1.8524045944213867, "learning_rate": 4.872205312265074e-07, "grpo_mean_advantage": -5.960464477539063e-08, "grpo_std_advantage": 3.460792754594877e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5601426362991333}
|
| 83 |
+
{"ts": "2025-12-27T21:40:38", "event": "train_log", "step": 410, "epoch": 1.6406406406406406, "progress_pct": 82.0, "epoch_pct": 82.03, "eta": "00:25:15", "max_grad_norm": 1.0, "loss": -0.5943, "grad_norm": 1.7269790172576904, "learning_rate": 4.3929438350970687e-07, "grpo_mean_advantage": 2.6226044269606064e-07, "grpo_std_advantage": 7.928817922220333e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.578656792640686}
|
| 84 |
+
{"ts": "2025-12-27T21:42:02", "event": "train_log", "step": 415, "epoch": 1.6606606606606606, "progress_pct": 83.0, "epoch_pct": 83.03, "eta": "00:23:51", "max_grad_norm": 1.0, "loss": -0.6193, "grad_norm": 2.26530122756958, "learning_rate": 3.936215871295634e-07, "grpo_mean_advantage": 2.3558736756967846e-06, "grpo_std_advantage": 1.4469559573626611e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5885810852050781}
|
| 85 |
+
{"ts": "2025-12-27T21:43:27", "event": "train_log", "step": 420, "epoch": 1.6806806806806807, "progress_pct": 84.0, "epoch_pct": 84.03, "eta": "00:22:27", "max_grad_norm": 1.0, "loss": -0.6934, "grad_norm": 2.6794464588165283, "learning_rate": 3.502520846183577e-07, "grpo_mean_advantage": 1.639127766850379e-08, "grpo_std_advantage": 9.352411325380672e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5805023312568665}
|
| 86 |
+
{"ts": "2025-12-27T21:44:52", "event": "train_log", "step": 425, "epoch": 1.7007007007007007, "progress_pct": 85.0, "epoch_pct": 85.04, "eta": "00:21:03", "max_grad_norm": 1.0, "loss": -0.5126, "grad_norm": 2.100447654724121, "learning_rate": 3.092332998903416e-07, "grpo_mean_advantage": 3.2387674764322583e-06, "grpo_std_advantage": 1.999079904635437e-05, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5655918121337891}
|
| 87 |
+
{"ts": "2025-12-27T21:46:17", "event": "train_log", "step": 430, "epoch": 1.7207207207207207, "progress_pct": 86.0, "epoch_pct": 86.04, "eta": "00:19:39", "max_grad_norm": 1.0, "loss": -0.5446, "grad_norm": 2.1027915477752686, "learning_rate": 2.706100863843822e-07, "grpo_mean_advantage": 3.5464762504489045e-07, "grpo_std_advantage": 1.7663603557593888e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5474504232406616}
|
| 88 |
+
{"ts": "2025-12-27T21:47:43", "event": "train_log", "step": 435, "epoch": 1.7407407407407407, "progress_pct": 87.0, "epoch_pct": 87.04, "eta": "00:18:15", "max_grad_norm": 1.0, "loss": -0.5125, "grad_norm": 2.289045572280884, "learning_rate": 2.3442467801738867e-07, "grpo_mean_advantage": 3.6135315895080566e-07, "grpo_std_advantage": 2.356920958845876e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5874254703521729}
|
| 89 |
+
{"ts": "2025-12-27T21:49:08", "event": "train_log", "step": 440, "epoch": 1.7607607607607607, "progress_pct": 88.0, "epoch_pct": 88.04, "eta": "00:16:51", "max_grad_norm": 1.0, "loss": -0.595, "grad_norm": 2.278038501739502, "learning_rate": 2.007166430021415e-07, "grpo_mean_advantage": 2.7567148563889532e-08, "grpo_std_advantage": 9.97340521280421e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5815118551254272}
|
| 90 |
+
{"ts": "2025-12-27T21:50:35", "event": "train_log", "step": 445, "epoch": 1.7807807807807807, "progress_pct": 89.0, "epoch_pct": 89.04, "eta": "00:15:27", "max_grad_norm": 1.0, "loss": -0.8055, "grad_norm": 2.340942621231079, "learning_rate": 1.6952284058003366e-07, "grpo_mean_advantage": -8.34465012644614e-08, "grpo_std_advantage": 5.558832185670326e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5611211061477661}
|
| 91 |
+
{"ts": "2025-12-27T21:52:06", "event": "train_log", "step": 450, "epoch": 1.800800800800801, "progress_pct": 90.0, "epoch_pct": 90.04, "eta": "00:14:03", "max_grad_norm": 1.0, "loss": -0.8561, "grad_norm": 2.4256298542022705, "learning_rate": 1.4087738071603075e-07, "grpo_mean_advantage": -1.9818544672034477e-07, "grpo_std_advantage": 6.800727305744658e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.590424656867981}
|
| 92 |
+
{"ts": "2025-12-27T21:53:32", "event": "train_log", "step": 455, "epoch": 1.820820820820821, "progress_pct": 91.0, "epoch_pct": 91.04, "eta": "00:12:39", "max_grad_norm": 1.0, "loss": -0.429, "grad_norm": 1.6453255414962769, "learning_rate": 1.1481158679992554e-07, "grpo_mean_advantage": -1.9371508841459217e-08, "grpo_std_advantage": 3.142378943721269e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5670351982116699}
|
| 93 |
+
{"ts": "2025-12-27T21:55:00", "event": "train_log", "step": 460, "epoch": 1.840840840840841, "progress_pct": 92.0, "epoch_pct": 92.04, "eta": "00:11:15", "max_grad_norm": 1.0, "loss": -0.642, "grad_norm": 2.3458049297332764, "learning_rate": 9.135396139467151e-08, "grpo_mean_advantage": 2.3692845729783585e-07, "grpo_std_advantage": 1.682946731307311e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5640432834625244}
|
| 94 |
+
{"ts": "2025-12-27T21:56:27", "event": "train_log", "step": 465, "epoch": 1.860860860860861, "progress_pct": 93.0, "epoch_pct": 93.04, "eta": "00:09:51", "max_grad_norm": 1.0, "loss": -0.583, "grad_norm": 2.730945110321045, "learning_rate": 7.053015506924749e-08, "grpo_mean_advantage": 1.110136480519941e-07, "grpo_std_advantage": 8.930008448260196e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5584251284599304}
|
| 95 |
+
{"ts": "2025-12-27T21:57:50", "event": "train_log", "step": 470, "epoch": 1.880880880880881, "progress_pct": 94.0, "epoch_pct": 94.04, "eta": "00:08:26", "max_grad_norm": 1.0, "loss": -0.5197, "grad_norm": 2.1463465690612793, "learning_rate": 5.236293835013839e-08, "grpo_mean_advantage": 2.5406478698641877e-07, "grpo_std_advantage": 9.93092498902115e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5392154455184937}
|
| 96 |
+
{"ts": "2025-12-27T21:59:15", "event": "train_log", "step": 475, "epoch": 1.900900900900901, "progress_pct": 95.0, "epoch_pct": 95.05, "eta": "00:07:02", "max_grad_norm": 1.0, "loss": -0.5864, "grad_norm": 2.427900791168213, "learning_rate": 3.687217682209837e-08, "grpo_mean_advantage": -8.940697071579962e-09, "grpo_std_advantage": 5.835169645251881e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5686308741569519}
|
| 97 |
+
{"ts": "2025-12-27T22:00:36", "event": "train_log", "step": 480, "epoch": 1.920920920920921, "progress_pct": 96.0, "epoch_pct": 96.05, "eta": "00:05:37", "max_grad_norm": 1.0, "loss": -0.5721, "grad_norm": 2.042795419692993, "learning_rate": 2.4074809405425227e-08, "grpo_mean_advantage": 4.0605664253234863e-07, "grpo_std_advantage": 2.3210795916384086e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5842767357826233}
|
| 98 |
+
{"ts": "2025-12-27T22:01:58", "event": "train_log", "step": 485, "epoch": 1.940940940940941, "progress_pct": 97.0, "epoch_pct": 97.05, "eta": "00:04:13", "max_grad_norm": 1.0, "loss": -0.5944, "grad_norm": 2.800136089324951, "learning_rate": 1.3984829833499636e-08, "grpo_mean_advantage": 1.341104507446289e-07, "grpo_std_advantage": 1.507950400991831e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5495311617851257}
|
| 99 |
+
{"ts": "2025-12-27T22:03:20", "event": "train_log", "step": 490, "epoch": 1.960960960960961, "progress_pct": 98.0, "epoch_pct": 98.05, "eta": "00:02:48", "max_grad_norm": 1.0, "loss": -0.7015, "grad_norm": 2.8475866317749023, "learning_rate": 6.6132713508446075e-09, "grpo_mean_advantage": 2.689659481802664e-07, "grpo_std_advantage": 8.491958851664094e-07, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.549436628818512}
|
| 100 |
+
{"ts": "2025-12-27T22:04:42", "event": "train_log", "step": 495, "epoch": 1.980980980980981, "progress_pct": 99.0, "epoch_pct": 99.05, "eta": "00:01:24", "max_grad_norm": 1.0, "loss": -0.4033, "grad_norm": 2.9422402381896973, "learning_rate": 1.9681946484320645e-09, "grpo_mean_advantage": 8.195638656616211e-08, "grpo_std_advantage": 3.802849732892355e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.544632077217102}
|
| 101 |
+
{"ts": "2025-12-27T22:06:01", "event": "train_log", "step": 500, "epoch": 2.0, "progress_pct": 100.0, "epoch_pct": 100.0, "eta": "00:00:00", "max_grad_norm": 1.0, "loss": -0.6773, "grad_norm": 2.66204833984375, "learning_rate": 5.467904943851077e-11, "grpo_mean_advantage": 7.552536089860951e-07, "grpo_std_advantage": 4.143997102801222e-06, "grpo_mean_kl_div": 0.0, "grpo_mean_group_score": 0.5968535542488098}
|
| 102 |
+
{"ts": "2025-12-27T22:06:03", "event": "train_log", "step": 500, "epoch": 2.0, "progress_pct": 100.0, "epoch_pct": 100.0, "eta": "00:00:00", "max_grad_norm": 1.0, "train_runtime": 8430.3461, "train_samples_per_second": 0.474, "train_steps_per_second": 0.059, "total_flos": 0.0, "train_loss": -0.23323759501613678}
|
grpo_qwen_14B_v2/wandb/debug-internal.log
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time":"2025-12-27T19:44:23.778699792Z","level":"INFO","msg":"stream: starting","core version":"0.23.1"}
|
| 2 |
+
{"time":"2025-12-27T19:44:23.931692267Z","level":"INFO","msg":"stream: created new stream","id":"jz7bptqa"}
|
| 3 |
+
{"time":"2025-12-27T19:44:23.931816408Z","level":"INFO","msg":"handler: started","stream_id":"jz7bptqa"}
|
| 4 |
+
{"time":"2025-12-27T19:44:23.931950499Z","level":"INFO","msg":"stream: started","id":"jz7bptqa"}
|
| 5 |
+
{"time":"2025-12-27T19:44:23.931981018Z","level":"INFO","msg":"writer: started","stream_id":"jz7bptqa"}
|
| 6 |
+
{"time":"2025-12-27T19:44:23.931980519Z","level":"INFO","msg":"sender: started","stream_id":"jz7bptqa"}
|
| 7 |
+
{"time":"2025-12-27T22:06:04.101056437Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
|
| 8 |
+
{"time":"2025-12-27T22:06:04.298272513Z","level":"INFO","msg":"handler: operation stats","stats":{}}
|
| 9 |
+
{"time":"2025-12-27T22:06:04.303656494Z","level":"INFO","msg":"stream: closing","id":"jz7bptqa"}
|
| 10 |
+
{"time":"2025-12-27T22:06:04.303677603Z","level":"INFO","msg":"handler: closed","stream_id":"jz7bptqa"}
|
| 11 |
+
{"time":"2025-12-27T22:06:04.303750712Z","level":"INFO","msg":"sender: closed","stream_id":"jz7bptqa"}
|
| 12 |
+
{"time":"2025-12-27T22:06:04.303767265Z","level":"INFO","msg":"stream: closed","id":"jz7bptqa"}
|
grpo_qwen_14B_v2/wandb/debug.log
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
2025-12-27 19:44:23,485 INFO MainThread:592889 [wandb_setup.py:_flush():80] Current SDK version is 0.23.1
|
| 2 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_setup.py:_flush():80] Configure stats pid to 592889
|
| 3 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_setup.py:_flush():80] Loading settings from /root/.config/wandb/settings
|
| 4 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_setup.py:_flush():80] Loading settings from /workspace/trainer-kit/GRPO-14b/wandb/settings
|
| 5 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_setup.py:_flush():80] Loading settings from environment variables
|
| 6 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_init.py:setup_run_log_directory():714] Logging user logs to runs/grpo_14b_run1/wandb/run-20251227_194423-jz7bptqa/logs/debug.log
|
| 7 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_init.py:setup_run_log_directory():715] Logging internal logs to runs/grpo_14b_run1/wandb/run-20251227_194423-jz7bptqa/logs/debug-internal.log
|
| 8 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_init.py:init():841] calling init triggers
|
| 9 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_init.py:init():846] wandb.init called with sweep_config: {}
|
| 10 |
+
config: {'model': {'repo_id': '/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2', 'tokenizer_name': 'Qwen/Qwen2.5-Coder-14B', 'load_in_8bit': False, 'load_in_4bit': False, 'torch_dtype': 'bfloat16', 'device_map': 'auto', 'trust_remote_code': True}, 'data': {'train_jsonl': 'grpo_dataset.jsonl', 'eval_jsonl': None, 'eval_split_ratio': 0.0, 'shuffle': True, 'num_proc': 1, 'prompt_field': 'prompt', 'completions_field': 'completions', 'scores_field': 'scores', 'format_type': 'raw', 'max_length': 2048, 'min_completions': 2, 'system_prompt': 'You are a Hyperswitch Rust code analyzer. Identify functions/structs that need modification for a given task.\n\n## Output Format\n\n##OUTPUT\nExplain the data flow and why each component must change:\n- Flow: [Input → Processing → Output with arrows]\n- For each component: "The [ComponentName] ([path]) must [action] because [reason]—without this, [consequence]"\n- Explain coupling between components\n\n##SELECT\nmodify::crates/path/to/file.rs::impl::ComponentName\nadd::crates/another/file.rs::function::AnotherComponent\n<EOS>\n\n## Rules\n\n1. Use full paths: `remove::crates/folder/file.rs::Type::Name`\n2. Use `::` for nested items: `status::StructName::Type::Name`\n3. Always explain "must change because" and "without this"\n3. Types of components: function, struct, enum, impl, trait\n4. If there is extra information (e.g., enum variants), include that too.\n5. Start with ##OUTPUT, end with ##SELECT, terminate with <EOS>\n\n## Example\n\n##TASK\nAdd webhook subscription support\n\n##OUTPUT\nThe webhook system routes events via EventClass enum. Flow: webhook → EventClass → handler → processing. The EventClass enum (crates/common_enums/src/enums.rs::EventClass) must add Subscriptions variant because it defines event routing—without this, subscription events cannot be processed. The SubscriptionStatus impl (crates/common_enums/src/transformers.rs::SubscriptionStatus) must map to EventType because it converts status to events—without this, status changes don\'t trigger webhooks. These are coupled: EventClass routes to handlers that use SubscriptionStatus mappings.\n\n##SELECT\ncrates/common_enums/src/enums.rs::EventClass\ncrates/common_enums/src/transformers.rs::SubscriptionStatus\n<EOS>\n', 'custom_template': '##INSTRUCTION\n{instruction}<|im_end|>\n{input}<|im_end|>\n{output}<|im_end|>'}, 'peft': {'enabled': True, 'r': 16, 'lora_alpha': 32, 'lora_dropout': 0.05, 'target_modules': ['q_proj', 'k_proj', 'v_proj', 'o_proj', 'gate_proj', 'up_proj', 'down_proj'], 'bias': 'none', 'task_type': 'CAUSAL_LM'}, 'grpo': {'group_size': 4, 'kl_coef': 0.05, 'normalize_advantages': True, 'reward_scaling': 1.0, 'reward_bias': 0.0, 'reward_clip': 5.0, 'advantage_temperature': 1.0, 'use_reference_model': False, 'seed': 42}, 'train': {'output_dir': 'runs/grpo_14b_run1', 'num_train_epochs': 2, 'per_device_train_batch_size': 1, 'gradient_accumulation_steps': 8, 'per_device_eval_batch_size': 1, 'learning_rate': 5e-06, 'weight_decay': 0.01, 'warmup_ratio': 0.05, 'lr_scheduler_type': 'cosine', 'fp16': False, 'bf16': True, 'max_grad_norm': 1.0, 'gradient_checkpointing': True, 'logging_steps': 5, 'save_steps': 100, 'save_total_limit': 2, 'evaluation_strategy': 'no', 'dataloader_num_workers': 4, 'dataloader_pin_memory': True, 'remove_unused_columns': False, 'report_to': [], 'seed': 42, 'ddp_find_unused_parameters': False}, 'run_dir': 'runs/grpo_14b_run1', '_wandb': {}}
|
| 11 |
+
2025-12-27 19:44:23,486 INFO MainThread:592889 [wandb_init.py:init():889] starting backend
|
| 12 |
+
2025-12-27 19:44:23,772 INFO MainThread:592889 [wandb_init.py:init():892] sending inform_init request
|
| 13 |
+
2025-12-27 19:44:23,776 INFO MainThread:592889 [wandb_init.py:init():900] backend started and connected
|
| 14 |
+
2025-12-27 19:44:23,778 INFO MainThread:592889 [wandb_init.py:init():970] updated telemetry
|
| 15 |
+
2025-12-27 19:44:23,779 INFO MainThread:592889 [wandb_init.py:init():994] communicating run to backend with 90.0 second timeout
|
| 16 |
+
2025-12-27 19:44:24,241 INFO MainThread:592889 [wandb_init.py:init():1041] starting run threads in backend
|
| 17 |
+
2025-12-27 19:44:24,354 INFO MainThread:592889 [wandb_run.py:_console_start():2521] atexit reg
|
| 18 |
+
2025-12-27 19:44:24,354 INFO MainThread:592889 [wandb_run.py:_redirect():2369] redirect: wrap_raw
|
| 19 |
+
2025-12-27 19:44:24,354 INFO MainThread:592889 [wandb_run.py:_redirect():2438] Wrapping output streams.
|
| 20 |
+
2025-12-27 19:44:24,354 INFO MainThread:592889 [wandb_run.py:_redirect():2461] Redirects installed.
|
| 21 |
+
2025-12-27 19:44:24,358 INFO MainThread:592889 [wandb_init.py:init():1081] run started, returning control to user process
|
| 22 |
+
2025-12-27 19:45:32,787 INFO MainThread:592889 [wandb_run.py:_config_callback():1396] config_cb None None {'peft_config': {'default': {'task_type': 'CAUSAL_LM', 'peft_type': 'LORA', 'auto_mapping': None, 'peft_version': '0.18.0', 'base_model_name_or_path': '/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2', 'revision': None, 'inference_mode': False, 'r': 16, 'target_modules': ['q_proj', 'o_proj', 'v_proj', 'up_proj', 'gate_proj', 'down_proj', 'k_proj'], 'exclude_modules': None, 'lora_alpha': 32, 'lora_dropout': 0.05, 'fan_in_fan_out': False, 'bias': 'none', 'use_rslora': False, 'modules_to_save': None, 'init_lora_weights': True, 'layers_to_transform': None, 'layers_pattern': None, 'rank_pattern': {}, 'alpha_pattern': {}, 'megatron_config': None, 'megatron_core': 'megatron.core', 'trainable_token_indices': None, 'loftq_config': {}, 'eva_config': None, 'corda_config': None, 'use_dora': False, 'alora_invocation_tokens': None, 'use_qalora': False, 'qalora_group_size': 16, 'layer_replication': None, 'runtime_config': {'ephemeral_gpu_offload': False}, 'lora_bias': False, 'target_parameters': None, 'arrow_config': None, 'ensure_weight_tying': False}}, 'vocab_size': 152064, 'max_position_embeddings': 32768, 'hidden_size': 5120, 'intermediate_size': 13824, 'num_hidden_layers': 48, 'num_attention_heads': 40, 'use_sliding_window': False, 'sliding_window': None, 'max_window_layers': 48, 'num_key_value_heads': 8, 'hidden_act': 'silu', 'initializer_range': 0.02, 'rms_norm_eps': 1e-06, 'use_cache': False, 'rope_theta': 10000.0, 'rope_scaling': None, 'attention_dropout': 0.0, 'layer_types': ['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'], 'return_dict': True, 'output_hidden_states': False, 'torchscript': False, 'dtype': 'bfloat16', 'pruned_heads': {}, 'tie_word_embeddings': False, 'chunk_size_feed_forward': 0, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'architectures': ['Qwen2ForCausalLM'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'task_specific_params': None, 'problem_type': None, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 151643, 'pad_token_id': None, 'eos_token_id': 151643, 'sep_token_id': None, 'decoder_start_token_id': None, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'suppress_tokens': None, 'begin_suppress_tokens': None, 'num_beam_groups': 1, 'diversity_penalty': 0.0, '_name_or_path': '/workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2', 'transformers_version': '4.57.3', 'model_type': 'qwen2', 'rope_parameters': {'rope_theta': 1000000.0, 'rope_type': 'default'}, 'tf_legacy_loss': False, 'use_bfloat16': False, 'output_attentions': False, 'output_dir': 'runs/grpo_14b_run1/checkpoints', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': 'no', 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 8, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 5e-06, 'weight_decay': 0.01, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 2.0, 'max_steps': -1, 'lr_scheduler_type': 'cosine', 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/grpo_14b_run1/checkpoints/runs/Dec27_19-45-32_a100-2gpu-shell-session-757d587799-mfdvv', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 5, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 100, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': 50, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': None, 'disable_tqdm': False, 'remove_unused_columns': False, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': 'eval_loss', 'greater_is_better': False, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}, 'parallelism_config': None, 'deepspeed': None, 'label_smoothing_factor': 0.0, 'optim': 'adamw_torch', 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'project': 'huggingface', 'trackio_space_id': 'trackio', 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': None, 'hub_model_id': None, 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'hub_private_repo': None, 'hub_always_push': False, 'hub_revision': None, 'gradient_checkpointing': False, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': 'no', 'neftune_noise_alpha': None, 'optim_target_modules': None, 'batch_eval_metrics': False, 'eval_on_start': False, 'use_liger_kernel': False, 'liger_kernel_config': None, 'eval_use_gather_object': False, 'average_tokens_across_devices': True}
|
| 23 |
+
2025-12-27 19:45:32,798 INFO MainThread:592889 [wandb_config.py:__setitem__():154] [no run ID] config set model/num_parameters = 14838846464 - <bound method Run._config_callback of <wandb.sdk.wandb_run.Run object at 0x7ada527aff70>>
|
| 24 |
+
2025-12-27 19:45:32,798 INFO MainThread:592889 [wandb_run.py:_config_callback():1396] config_cb model/num_parameters 14838846464 None
|
| 25 |
+
2025-12-27 22:06:03,800 INFO MainThread:592889 [wandb_run.py:_finish():2287] finishing run sirajuddin-shaik-007/rl-training/jz7bptqa
|
| 26 |
+
2025-12-27 22:06:03,800 INFO MainThread:592889 [wandb_run.py:_atexit_cleanup():2486] got exitcode: 0
|
| 27 |
+
2025-12-27 22:06:03,801 INFO MainThread:592889 [wandb_run.py:_restore():2468] restore
|
| 28 |
+
2025-12-27 22:06:03,801 INFO MainThread:592889 [wandb_run.py:_restore():2474] restore done
|
| 29 |
+
2025-12-27 22:06:04,302 INFO MainThread:592889 [wandb_run.py:_footer_sync_info():3862] logging synced files
|
grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/files/config.yaml
ADDED
|
@@ -0,0 +1,725 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_name_or_path:
|
| 2 |
+
value: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 3 |
+
_wandb:
|
| 4 |
+
value:
|
| 5 |
+
cli_version: 0.23.1
|
| 6 |
+
e:
|
| 7 |
+
flq59s27gswy5t5lhj0bnmni5u7y1dm1:
|
| 8 |
+
args:
|
| 9 |
+
- --config
|
| 10 |
+
- config_grpo_14b.yaml
|
| 11 |
+
codePath: run_grpo_fixed.py
|
| 12 |
+
codePathLocal: run_grpo_fixed.py
|
| 13 |
+
cpu_count: 12
|
| 14 |
+
cpu_count_logical: 24
|
| 15 |
+
cudaVersion: "13.0"
|
| 16 |
+
disk:
|
| 17 |
+
/:
|
| 18 |
+
total: "791251738624"
|
| 19 |
+
used: "537236914176"
|
| 20 |
+
email: [email protected]
|
| 21 |
+
executable: /root/llm_finetuning_env/bin/python
|
| 22 |
+
gpu: NVIDIA A100-SXM4-80GB
|
| 23 |
+
gpu_count: 2
|
| 24 |
+
gpu_nvidia:
|
| 25 |
+
- architecture: Ampere
|
| 26 |
+
cudaCores: 6912
|
| 27 |
+
memoryTotal: "85899345920"
|
| 28 |
+
name: NVIDIA A100-SXM4-80GB
|
| 29 |
+
uuid: GPU-989794b0-ec3b-13bf-db9f-3fbe341497ba
|
| 30 |
+
- architecture: Ampere
|
| 31 |
+
cudaCores: 6912
|
| 32 |
+
memoryTotal: "85899345920"
|
| 33 |
+
name: NVIDIA A100-SXM4-80GB
|
| 34 |
+
uuid: GPU-3790aa64-60ef-9eac-b0b1-b278ee8c0d40
|
| 35 |
+
host: a100-2gpu-shell-session-757d587799-mfdvv
|
| 36 |
+
memory:
|
| 37 |
+
total: "359047892992"
|
| 38 |
+
os: Linux-6.12.46+-x86_64-with-glibc2.35
|
| 39 |
+
program: /workspace/trainer-kit/GRPO-14b/run_grpo_fixed.py
|
| 40 |
+
python: CPython 3.10.12
|
| 41 |
+
root: runs/grpo_14b_run1
|
| 42 |
+
startedAt: "2025-12-27T19:44:23.484378Z"
|
| 43 |
+
writerId: flq59s27gswy5t5lhj0bnmni5u7y1dm1
|
| 44 |
+
m:
|
| 45 |
+
- "1": train/global_step
|
| 46 |
+
"6":
|
| 47 |
+
- 3
|
| 48 |
+
"7": []
|
| 49 |
+
- "2": '*'
|
| 50 |
+
"5": 1
|
| 51 |
+
"6":
|
| 52 |
+
- 1
|
| 53 |
+
"7": []
|
| 54 |
+
python_version: 3.10.12
|
| 55 |
+
t:
|
| 56 |
+
"1":
|
| 57 |
+
- 1
|
| 58 |
+
- 11
|
| 59 |
+
- 41
|
| 60 |
+
- 49
|
| 61 |
+
- 51
|
| 62 |
+
- 71
|
| 63 |
+
- 98
|
| 64 |
+
"2":
|
| 65 |
+
- 1
|
| 66 |
+
- 11
|
| 67 |
+
- 41
|
| 68 |
+
- 49
|
| 69 |
+
- 51
|
| 70 |
+
- 71
|
| 71 |
+
- 98
|
| 72 |
+
"3":
|
| 73 |
+
- 2
|
| 74 |
+
- 7
|
| 75 |
+
- 15
|
| 76 |
+
- 16
|
| 77 |
+
- 19
|
| 78 |
+
- 62
|
| 79 |
+
- 66
|
| 80 |
+
"4": 3.10.12
|
| 81 |
+
"5": 0.23.1
|
| 82 |
+
"6": 4.57.3
|
| 83 |
+
"9":
|
| 84 |
+
"1": transformers_trainer
|
| 85 |
+
"12": 0.23.1
|
| 86 |
+
"13": linux-x86_64
|
| 87 |
+
accelerator_config:
|
| 88 |
+
value:
|
| 89 |
+
dispatch_batches: null
|
| 90 |
+
even_batches: true
|
| 91 |
+
gradient_accumulation_kwargs: null
|
| 92 |
+
non_blocking: false
|
| 93 |
+
split_batches: false
|
| 94 |
+
use_seedable_sampler: true
|
| 95 |
+
adafactor:
|
| 96 |
+
value: false
|
| 97 |
+
adam_beta1:
|
| 98 |
+
value: 0.9
|
| 99 |
+
adam_beta2:
|
| 100 |
+
value: 0.999
|
| 101 |
+
adam_epsilon:
|
| 102 |
+
value: 1e-08
|
| 103 |
+
add_cross_attention:
|
| 104 |
+
value: false
|
| 105 |
+
architectures:
|
| 106 |
+
value:
|
| 107 |
+
- Qwen2ForCausalLM
|
| 108 |
+
attention_dropout:
|
| 109 |
+
value: 0
|
| 110 |
+
auto_find_batch_size:
|
| 111 |
+
value: false
|
| 112 |
+
average_tokens_across_devices:
|
| 113 |
+
value: true
|
| 114 |
+
bad_words_ids:
|
| 115 |
+
value: null
|
| 116 |
+
batch_eval_metrics:
|
| 117 |
+
value: false
|
| 118 |
+
begin_suppress_tokens:
|
| 119 |
+
value: null
|
| 120 |
+
bf16:
|
| 121 |
+
value: true
|
| 122 |
+
bf16_full_eval:
|
| 123 |
+
value: false
|
| 124 |
+
bos_token_id:
|
| 125 |
+
value: 151643
|
| 126 |
+
chunk_size_feed_forward:
|
| 127 |
+
value: 0
|
| 128 |
+
cross_attention_hidden_size:
|
| 129 |
+
value: null
|
| 130 |
+
data:
|
| 131 |
+
value:
|
| 132 |
+
completions_field: completions
|
| 133 |
+
custom_template: |-
|
| 134 |
+
##INSTRUCTION
|
| 135 |
+
{instruction}<|im_end|>
|
| 136 |
+
{input}<|im_end|>
|
| 137 |
+
{output}<|im_end|>
|
| 138 |
+
eval_jsonl: null
|
| 139 |
+
eval_split_ratio: 0
|
| 140 |
+
format_type: raw
|
| 141 |
+
max_length: 2048
|
| 142 |
+
min_completions: 2
|
| 143 |
+
num_proc: 1
|
| 144 |
+
prompt_field: prompt
|
| 145 |
+
scores_field: scores
|
| 146 |
+
shuffle: true
|
| 147 |
+
system_prompt: |
|
| 148 |
+
You are a Hyperswitch Rust code analyzer. Identify functions/structs that need modification for a given task.
|
| 149 |
+
|
| 150 |
+
## Output Format
|
| 151 |
+
|
| 152 |
+
##OUTPUT
|
| 153 |
+
Explain the data flow and why each component must change:
|
| 154 |
+
- Flow: [Input → Processing → Output with arrows]
|
| 155 |
+
- For each component: "The [ComponentName] ([path]) must [action] because [reason]—without this, [consequence]"
|
| 156 |
+
- Explain coupling between components
|
| 157 |
+
|
| 158 |
+
##SELECT
|
| 159 |
+
modify::crates/path/to/file.rs::impl::ComponentName
|
| 160 |
+
add::crates/another/file.rs::function::AnotherComponent
|
| 161 |
+
<EOS>
|
| 162 |
+
|
| 163 |
+
## Rules
|
| 164 |
+
|
| 165 |
+
1. Use full paths: `remove::crates/folder/file.rs::Type::Name`
|
| 166 |
+
2. Use `::` for nested items: `status::StructName::Type::Name`
|
| 167 |
+
3. Always explain "must change because" and "without this"
|
| 168 |
+
3. Types of components: function, struct, enum, impl, trait
|
| 169 |
+
4. If there is extra information (e.g., enum variants), include that too.
|
| 170 |
+
5. Start with ##OUTPUT, end with ##SELECT, terminate with <EOS>
|
| 171 |
+
|
| 172 |
+
## Example
|
| 173 |
+
|
| 174 |
+
##TASK
|
| 175 |
+
Add webhook subscription support
|
| 176 |
+
|
| 177 |
+
##OUTPUT
|
| 178 |
+
The webhook system routes events via EventClass enum. Flow: webhook → EventClass → handler → processing. The EventClass enum (crates/common_enums/src/enums.rs::EventClass) must add Subscriptions variant because it defines event routing—without this, subscription events cannot be processed. The SubscriptionStatus impl (crates/common_enums/src/transformers.rs::SubscriptionStatus) must map to EventType because it converts status to events—without this, status changes don't trigger webhooks. These are coupled: EventClass routes to handlers that use SubscriptionStatus mappings.
|
| 179 |
+
|
| 180 |
+
##SELECT
|
| 181 |
+
crates/common_enums/src/enums.rs::EventClass
|
| 182 |
+
crates/common_enums/src/transformers.rs::SubscriptionStatus
|
| 183 |
+
<EOS>
|
| 184 |
+
train_jsonl: grpo_dataset.jsonl
|
| 185 |
+
data_seed:
|
| 186 |
+
value: null
|
| 187 |
+
dataloader_drop_last:
|
| 188 |
+
value: false
|
| 189 |
+
dataloader_num_workers:
|
| 190 |
+
value: 0
|
| 191 |
+
dataloader_persistent_workers:
|
| 192 |
+
value: false
|
| 193 |
+
dataloader_pin_memory:
|
| 194 |
+
value: true
|
| 195 |
+
dataloader_prefetch_factor:
|
| 196 |
+
value: null
|
| 197 |
+
ddp_backend:
|
| 198 |
+
value: null
|
| 199 |
+
ddp_broadcast_buffers:
|
| 200 |
+
value: null
|
| 201 |
+
ddp_bucket_cap_mb:
|
| 202 |
+
value: null
|
| 203 |
+
ddp_find_unused_parameters:
|
| 204 |
+
value: null
|
| 205 |
+
ddp_timeout:
|
| 206 |
+
value: 1800
|
| 207 |
+
debug:
|
| 208 |
+
value: []
|
| 209 |
+
decoder_start_token_id:
|
| 210 |
+
value: null
|
| 211 |
+
deepspeed:
|
| 212 |
+
value: null
|
| 213 |
+
disable_tqdm:
|
| 214 |
+
value: false
|
| 215 |
+
diversity_penalty:
|
| 216 |
+
value: 0
|
| 217 |
+
do_eval:
|
| 218 |
+
value: false
|
| 219 |
+
do_predict:
|
| 220 |
+
value: false
|
| 221 |
+
do_sample:
|
| 222 |
+
value: false
|
| 223 |
+
do_train:
|
| 224 |
+
value: false
|
| 225 |
+
dtype:
|
| 226 |
+
value: bfloat16
|
| 227 |
+
early_stopping:
|
| 228 |
+
value: false
|
| 229 |
+
encoder_no_repeat_ngram_size:
|
| 230 |
+
value: 0
|
| 231 |
+
eos_token_id:
|
| 232 |
+
value: 151643
|
| 233 |
+
eval_accumulation_steps:
|
| 234 |
+
value: null
|
| 235 |
+
eval_delay:
|
| 236 |
+
value: 0
|
| 237 |
+
eval_do_concat_batches:
|
| 238 |
+
value: true
|
| 239 |
+
eval_on_start:
|
| 240 |
+
value: false
|
| 241 |
+
eval_steps:
|
| 242 |
+
value: 50
|
| 243 |
+
eval_strategy:
|
| 244 |
+
value: "no"
|
| 245 |
+
eval_use_gather_object:
|
| 246 |
+
value: false
|
| 247 |
+
exponential_decay_length_penalty:
|
| 248 |
+
value: null
|
| 249 |
+
finetuning_task:
|
| 250 |
+
value: null
|
| 251 |
+
forced_bos_token_id:
|
| 252 |
+
value: null
|
| 253 |
+
forced_eos_token_id:
|
| 254 |
+
value: null
|
| 255 |
+
fp16:
|
| 256 |
+
value: false
|
| 257 |
+
fp16_backend:
|
| 258 |
+
value: auto
|
| 259 |
+
fp16_full_eval:
|
| 260 |
+
value: false
|
| 261 |
+
fp16_opt_level:
|
| 262 |
+
value: O1
|
| 263 |
+
fsdp:
|
| 264 |
+
value: []
|
| 265 |
+
fsdp_config:
|
| 266 |
+
value:
|
| 267 |
+
min_num_params: 0
|
| 268 |
+
xla: false
|
| 269 |
+
xla_fsdp_grad_ckpt: false
|
| 270 |
+
xla_fsdp_v2: false
|
| 271 |
+
fsdp_min_num_params:
|
| 272 |
+
value: 0
|
| 273 |
+
fsdp_transformer_layer_cls_to_wrap:
|
| 274 |
+
value: null
|
| 275 |
+
full_determinism:
|
| 276 |
+
value: false
|
| 277 |
+
gradient_accumulation_steps:
|
| 278 |
+
value: 8
|
| 279 |
+
gradient_checkpointing:
|
| 280 |
+
value: false
|
| 281 |
+
gradient_checkpointing_kwargs:
|
| 282 |
+
value: null
|
| 283 |
+
greater_is_better:
|
| 284 |
+
value: false
|
| 285 |
+
group_by_length:
|
| 286 |
+
value: false
|
| 287 |
+
grpo:
|
| 288 |
+
value:
|
| 289 |
+
advantage_temperature: 1
|
| 290 |
+
group_size: 4
|
| 291 |
+
kl_coef: 0.05
|
| 292 |
+
normalize_advantages: true
|
| 293 |
+
reward_bias: 0
|
| 294 |
+
reward_clip: 5
|
| 295 |
+
reward_scaling: 1
|
| 296 |
+
seed: 42
|
| 297 |
+
use_reference_model: false
|
| 298 |
+
half_precision_backend:
|
| 299 |
+
value: auto
|
| 300 |
+
hidden_act:
|
| 301 |
+
value: silu
|
| 302 |
+
hidden_size:
|
| 303 |
+
value: 5120
|
| 304 |
+
hub_always_push:
|
| 305 |
+
value: false
|
| 306 |
+
hub_model_id:
|
| 307 |
+
value: null
|
| 308 |
+
hub_private_repo:
|
| 309 |
+
value: null
|
| 310 |
+
hub_revision:
|
| 311 |
+
value: null
|
| 312 |
+
hub_strategy:
|
| 313 |
+
value: every_save
|
| 314 |
+
hub_token:
|
| 315 |
+
value: <HUB_TOKEN>
|
| 316 |
+
id2label:
|
| 317 |
+
value:
|
| 318 |
+
"0": LABEL_0
|
| 319 |
+
"1": LABEL_1
|
| 320 |
+
ignore_data_skip:
|
| 321 |
+
value: false
|
| 322 |
+
include_for_metrics:
|
| 323 |
+
value: []
|
| 324 |
+
include_inputs_for_metrics:
|
| 325 |
+
value: false
|
| 326 |
+
include_num_input_tokens_seen:
|
| 327 |
+
value: "no"
|
| 328 |
+
include_tokens_per_second:
|
| 329 |
+
value: false
|
| 330 |
+
initializer_range:
|
| 331 |
+
value: 0.02
|
| 332 |
+
intermediate_size:
|
| 333 |
+
value: 13824
|
| 334 |
+
is_decoder:
|
| 335 |
+
value: false
|
| 336 |
+
is_encoder_decoder:
|
| 337 |
+
value: false
|
| 338 |
+
jit_mode_eval:
|
| 339 |
+
value: false
|
| 340 |
+
label_names:
|
| 341 |
+
value: null
|
| 342 |
+
label_smoothing_factor:
|
| 343 |
+
value: 0
|
| 344 |
+
label2id:
|
| 345 |
+
value:
|
| 346 |
+
LABEL_0: 0
|
| 347 |
+
LABEL_1: 1
|
| 348 |
+
layer_types:
|
| 349 |
+
value:
|
| 350 |
+
- full_attention
|
| 351 |
+
- full_attention
|
| 352 |
+
- full_attention
|
| 353 |
+
- full_attention
|
| 354 |
+
- full_attention
|
| 355 |
+
- full_attention
|
| 356 |
+
- full_attention
|
| 357 |
+
- full_attention
|
| 358 |
+
- full_attention
|
| 359 |
+
- full_attention
|
| 360 |
+
- full_attention
|
| 361 |
+
- full_attention
|
| 362 |
+
- full_attention
|
| 363 |
+
- full_attention
|
| 364 |
+
- full_attention
|
| 365 |
+
- full_attention
|
| 366 |
+
- full_attention
|
| 367 |
+
- full_attention
|
| 368 |
+
- full_attention
|
| 369 |
+
- full_attention
|
| 370 |
+
- full_attention
|
| 371 |
+
- full_attention
|
| 372 |
+
- full_attention
|
| 373 |
+
- full_attention
|
| 374 |
+
- full_attention
|
| 375 |
+
- full_attention
|
| 376 |
+
- full_attention
|
| 377 |
+
- full_attention
|
| 378 |
+
- full_attention
|
| 379 |
+
- full_attention
|
| 380 |
+
- full_attention
|
| 381 |
+
- full_attention
|
| 382 |
+
- full_attention
|
| 383 |
+
- full_attention
|
| 384 |
+
- full_attention
|
| 385 |
+
- full_attention
|
| 386 |
+
- full_attention
|
| 387 |
+
- full_attention
|
| 388 |
+
- full_attention
|
| 389 |
+
- full_attention
|
| 390 |
+
- full_attention
|
| 391 |
+
- full_attention
|
| 392 |
+
- full_attention
|
| 393 |
+
- full_attention
|
| 394 |
+
- full_attention
|
| 395 |
+
- full_attention
|
| 396 |
+
- full_attention
|
| 397 |
+
- full_attention
|
| 398 |
+
learning_rate:
|
| 399 |
+
value: 5e-06
|
| 400 |
+
length_column_name:
|
| 401 |
+
value: length
|
| 402 |
+
length_penalty:
|
| 403 |
+
value: 1
|
| 404 |
+
liger_kernel_config:
|
| 405 |
+
value: null
|
| 406 |
+
load_best_model_at_end:
|
| 407 |
+
value: false
|
| 408 |
+
local_rank:
|
| 409 |
+
value: 0
|
| 410 |
+
log_level:
|
| 411 |
+
value: passive
|
| 412 |
+
log_level_replica:
|
| 413 |
+
value: warning
|
| 414 |
+
log_on_each_node:
|
| 415 |
+
value: true
|
| 416 |
+
logging_dir:
|
| 417 |
+
value: runs/grpo_14b_run1/checkpoints/runs/Dec27_19-45-32_a100-2gpu-shell-session-757d587799-mfdvv
|
| 418 |
+
logging_first_step:
|
| 419 |
+
value: false
|
| 420 |
+
logging_nan_inf_filter:
|
| 421 |
+
value: true
|
| 422 |
+
logging_steps:
|
| 423 |
+
value: 5
|
| 424 |
+
logging_strategy:
|
| 425 |
+
value: steps
|
| 426 |
+
lr_scheduler_type:
|
| 427 |
+
value: cosine
|
| 428 |
+
max_grad_norm:
|
| 429 |
+
value: 1
|
| 430 |
+
max_length:
|
| 431 |
+
value: 20
|
| 432 |
+
max_position_embeddings:
|
| 433 |
+
value: 32768
|
| 434 |
+
max_steps:
|
| 435 |
+
value: -1
|
| 436 |
+
max_window_layers:
|
| 437 |
+
value: 48
|
| 438 |
+
metric_for_best_model:
|
| 439 |
+
value: eval_loss
|
| 440 |
+
min_length:
|
| 441 |
+
value: 0
|
| 442 |
+
model:
|
| 443 |
+
value:
|
| 444 |
+
device_map: auto
|
| 445 |
+
load_in_4bit: false
|
| 446 |
+
load_in_8bit: false
|
| 447 |
+
repo_id: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 448 |
+
tokenizer_name: Qwen/Qwen2.5-Coder-14B
|
| 449 |
+
torch_dtype: bfloat16
|
| 450 |
+
trust_remote_code: true
|
| 451 |
+
model/num_parameters:
|
| 452 |
+
value: 14838846464
|
| 453 |
+
model_type:
|
| 454 |
+
value: qwen2
|
| 455 |
+
mp_parameters:
|
| 456 |
+
value: ""
|
| 457 |
+
neftune_noise_alpha:
|
| 458 |
+
value: null
|
| 459 |
+
no_cuda:
|
| 460 |
+
value: false
|
| 461 |
+
no_repeat_ngram_size:
|
| 462 |
+
value: 0
|
| 463 |
+
num_attention_heads:
|
| 464 |
+
value: 40
|
| 465 |
+
num_beam_groups:
|
| 466 |
+
value: 1
|
| 467 |
+
num_beams:
|
| 468 |
+
value: 1
|
| 469 |
+
num_hidden_layers:
|
| 470 |
+
value: 48
|
| 471 |
+
num_key_value_heads:
|
| 472 |
+
value: 8
|
| 473 |
+
num_return_sequences:
|
| 474 |
+
value: 1
|
| 475 |
+
num_train_epochs:
|
| 476 |
+
value: 2
|
| 477 |
+
optim:
|
| 478 |
+
value: adamw_torch
|
| 479 |
+
optim_args:
|
| 480 |
+
value: null
|
| 481 |
+
optim_target_modules:
|
| 482 |
+
value: null
|
| 483 |
+
output_attentions:
|
| 484 |
+
value: false
|
| 485 |
+
output_dir:
|
| 486 |
+
value: runs/grpo_14b_run1/checkpoints
|
| 487 |
+
output_hidden_states:
|
| 488 |
+
value: false
|
| 489 |
+
output_scores:
|
| 490 |
+
value: false
|
| 491 |
+
overwrite_output_dir:
|
| 492 |
+
value: false
|
| 493 |
+
pad_token_id:
|
| 494 |
+
value: null
|
| 495 |
+
parallelism_config:
|
| 496 |
+
value: null
|
| 497 |
+
past_index:
|
| 498 |
+
value: -1
|
| 499 |
+
peft:
|
| 500 |
+
value:
|
| 501 |
+
bias: none
|
| 502 |
+
enabled: true
|
| 503 |
+
lora_alpha: 32
|
| 504 |
+
lora_dropout: 0.05
|
| 505 |
+
r: 16
|
| 506 |
+
target_modules:
|
| 507 |
+
- q_proj
|
| 508 |
+
- k_proj
|
| 509 |
+
- v_proj
|
| 510 |
+
- o_proj
|
| 511 |
+
- gate_proj
|
| 512 |
+
- up_proj
|
| 513 |
+
- down_proj
|
| 514 |
+
task_type: CAUSAL_LM
|
| 515 |
+
peft_config:
|
| 516 |
+
value:
|
| 517 |
+
default:
|
| 518 |
+
alora_invocation_tokens: null
|
| 519 |
+
arrow_config: null
|
| 520 |
+
auto_mapping: null
|
| 521 |
+
base_model_name_or_path: /workspace/Models/Qwen2.5-Coder-14B-CPT-SFT_v2
|
| 522 |
+
bias: none
|
| 523 |
+
corda_config: null
|
| 524 |
+
ensure_weight_tying: false
|
| 525 |
+
eva_config: null
|
| 526 |
+
exclude_modules: null
|
| 527 |
+
fan_in_fan_out: false
|
| 528 |
+
inference_mode: false
|
| 529 |
+
init_lora_weights: true
|
| 530 |
+
layer_replication: null
|
| 531 |
+
layers_pattern: null
|
| 532 |
+
layers_to_transform: null
|
| 533 |
+
lora_alpha: 32
|
| 534 |
+
lora_bias: false
|
| 535 |
+
lora_dropout: 0.05
|
| 536 |
+
megatron_config: null
|
| 537 |
+
megatron_core: megatron.core
|
| 538 |
+
modules_to_save: null
|
| 539 |
+
peft_type: LORA
|
| 540 |
+
peft_version: 0.18.0
|
| 541 |
+
qalora_group_size: 16
|
| 542 |
+
r: 16
|
| 543 |
+
revision: null
|
| 544 |
+
runtime_config:
|
| 545 |
+
ephemeral_gpu_offload: false
|
| 546 |
+
target_modules:
|
| 547 |
+
- q_proj
|
| 548 |
+
- o_proj
|
| 549 |
+
- v_proj
|
| 550 |
+
- up_proj
|
| 551 |
+
- gate_proj
|
| 552 |
+
- down_proj
|
| 553 |
+
- k_proj
|
| 554 |
+
target_parameters: null
|
| 555 |
+
task_type: CAUSAL_LM
|
| 556 |
+
trainable_token_indices: null
|
| 557 |
+
use_dora: false
|
| 558 |
+
use_qalora: false
|
| 559 |
+
use_rslora: false
|
| 560 |
+
per_device_eval_batch_size:
|
| 561 |
+
value: 1
|
| 562 |
+
per_device_train_batch_size:
|
| 563 |
+
value: 1
|
| 564 |
+
per_gpu_eval_batch_size:
|
| 565 |
+
value: null
|
| 566 |
+
per_gpu_train_batch_size:
|
| 567 |
+
value: null
|
| 568 |
+
prediction_loss_only:
|
| 569 |
+
value: false
|
| 570 |
+
prefix:
|
| 571 |
+
value: null
|
| 572 |
+
problem_type:
|
| 573 |
+
value: null
|
| 574 |
+
project:
|
| 575 |
+
value: huggingface
|
| 576 |
+
push_to_hub:
|
| 577 |
+
value: false
|
| 578 |
+
push_to_hub_model_id:
|
| 579 |
+
value: null
|
| 580 |
+
push_to_hub_organization:
|
| 581 |
+
value: null
|
| 582 |
+
push_to_hub_token:
|
| 583 |
+
value: <PUSH_TO_HUB_TOKEN>
|
| 584 |
+
ray_scope:
|
| 585 |
+
value: last
|
| 586 |
+
remove_invalid_values:
|
| 587 |
+
value: false
|
| 588 |
+
remove_unused_columns:
|
| 589 |
+
value: false
|
| 590 |
+
repetition_penalty:
|
| 591 |
+
value: 1
|
| 592 |
+
report_to:
|
| 593 |
+
value:
|
| 594 |
+
- wandb
|
| 595 |
+
restore_callback_states_from_checkpoint:
|
| 596 |
+
value: false
|
| 597 |
+
resume_from_checkpoint:
|
| 598 |
+
value: null
|
| 599 |
+
return_dict:
|
| 600 |
+
value: true
|
| 601 |
+
return_dict_in_generate:
|
| 602 |
+
value: false
|
| 603 |
+
rms_norm_eps:
|
| 604 |
+
value: 1e-06
|
| 605 |
+
rope_parameters:
|
| 606 |
+
value:
|
| 607 |
+
rope_theta: 1e+06
|
| 608 |
+
rope_type: default
|
| 609 |
+
rope_scaling:
|
| 610 |
+
value: null
|
| 611 |
+
rope_theta:
|
| 612 |
+
value: 10000
|
| 613 |
+
run_dir:
|
| 614 |
+
value: runs/grpo_14b_run1
|
| 615 |
+
run_name:
|
| 616 |
+
value: null
|
| 617 |
+
save_on_each_node:
|
| 618 |
+
value: false
|
| 619 |
+
save_only_model:
|
| 620 |
+
value: false
|
| 621 |
+
save_safetensors:
|
| 622 |
+
value: true
|
| 623 |
+
save_steps:
|
| 624 |
+
value: 100
|
| 625 |
+
save_strategy:
|
| 626 |
+
value: steps
|
| 627 |
+
save_total_limit:
|
| 628 |
+
value: 2
|
| 629 |
+
seed:
|
| 630 |
+
value: 42
|
| 631 |
+
sep_token_id:
|
| 632 |
+
value: null
|
| 633 |
+
skip_memory_metrics:
|
| 634 |
+
value: true
|
| 635 |
+
sliding_window:
|
| 636 |
+
value: null
|
| 637 |
+
suppress_tokens:
|
| 638 |
+
value: null
|
| 639 |
+
task_specific_params:
|
| 640 |
+
value: null
|
| 641 |
+
temperature:
|
| 642 |
+
value: 1
|
| 643 |
+
tf_legacy_loss:
|
| 644 |
+
value: false
|
| 645 |
+
tf32:
|
| 646 |
+
value: null
|
| 647 |
+
tie_encoder_decoder:
|
| 648 |
+
value: false
|
| 649 |
+
tie_word_embeddings:
|
| 650 |
+
value: false
|
| 651 |
+
tokenizer_class:
|
| 652 |
+
value: null
|
| 653 |
+
top_k:
|
| 654 |
+
value: 50
|
| 655 |
+
top_p:
|
| 656 |
+
value: 1
|
| 657 |
+
torch_compile:
|
| 658 |
+
value: false
|
| 659 |
+
torch_compile_backend:
|
| 660 |
+
value: null
|
| 661 |
+
torch_compile_mode:
|
| 662 |
+
value: null
|
| 663 |
+
torch_empty_cache_steps:
|
| 664 |
+
value: null
|
| 665 |
+
torchdynamo:
|
| 666 |
+
value: null
|
| 667 |
+
torchscript:
|
| 668 |
+
value: false
|
| 669 |
+
tpu_metrics_debug:
|
| 670 |
+
value: false
|
| 671 |
+
tpu_num_cores:
|
| 672 |
+
value: null
|
| 673 |
+
trackio_space_id:
|
| 674 |
+
value: trackio
|
| 675 |
+
train:
|
| 676 |
+
value:
|
| 677 |
+
bf16: true
|
| 678 |
+
dataloader_num_workers: 4
|
| 679 |
+
dataloader_pin_memory: true
|
| 680 |
+
ddp_find_unused_parameters: false
|
| 681 |
+
evaluation_strategy: "no"
|
| 682 |
+
fp16: false
|
| 683 |
+
gradient_accumulation_steps: 8
|
| 684 |
+
gradient_checkpointing: true
|
| 685 |
+
learning_rate: 5e-06
|
| 686 |
+
logging_steps: 5
|
| 687 |
+
lr_scheduler_type: cosine
|
| 688 |
+
max_grad_norm: 1
|
| 689 |
+
num_train_epochs: 2
|
| 690 |
+
output_dir: runs/grpo_14b_run1
|
| 691 |
+
per_device_eval_batch_size: 1
|
| 692 |
+
per_device_train_batch_size: 1
|
| 693 |
+
remove_unused_columns: false
|
| 694 |
+
report_to: []
|
| 695 |
+
save_steps: 100
|
| 696 |
+
save_total_limit: 2
|
| 697 |
+
seed: 42
|
| 698 |
+
warmup_ratio: 0.05
|
| 699 |
+
weight_decay: 0.01
|
| 700 |
+
transformers_version:
|
| 701 |
+
value: 4.57.3
|
| 702 |
+
typical_p:
|
| 703 |
+
value: 1
|
| 704 |
+
use_bfloat16:
|
| 705 |
+
value: false
|
| 706 |
+
use_cache:
|
| 707 |
+
value: false
|
| 708 |
+
use_cpu:
|
| 709 |
+
value: false
|
| 710 |
+
use_legacy_prediction_loop:
|
| 711 |
+
value: false
|
| 712 |
+
use_liger_kernel:
|
| 713 |
+
value: false
|
| 714 |
+
use_mps_device:
|
| 715 |
+
value: false
|
| 716 |
+
use_sliding_window:
|
| 717 |
+
value: false
|
| 718 |
+
vocab_size:
|
| 719 |
+
value: 152064
|
| 720 |
+
warmup_ratio:
|
| 721 |
+
value: 0.05
|
| 722 |
+
warmup_steps:
|
| 723 |
+
value: 0
|
| 724 |
+
weight_decay:
|
| 725 |
+
value: 0.01
|
grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/files/output.log
ADDED
|
@@ -0,0 +1,134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Wandb initialized: project='rl-training', name='auto-generated'
|
| 2 |
+
`torch_dtype` is deprecated! Use `dtype` instead!
|
| 3 |
+
Loading checkpoint shards: 100%|████████████████████████████████████████████| 16/16 [00:17<00:00, 1.10s/it]
|
| 4 |
+
Formatting and tokenizing train GRPO data (num_proc=1): 100%|████| 2000/2000 [00:37<00:00, 53.65 examples/s]
|
| 5 |
+
Filter: 100%|███████████████████████████████████████████████████| 1998/1998 [00:08<00:00, 244.33 examples/s]
|
| 6 |
+
Training samples: 1998
|
| 7 |
+
Warning: group_size (4) > min_completions (2)
|
| 8 |
+
Setting group_size to 2
|
| 9 |
+
|
| 10 |
+
=== GRPO Training Configuration ===
|
| 11 |
+
Group size: 2
|
| 12 |
+
KL coefficient: 0.05
|
| 13 |
+
Normalize advantages: True
|
| 14 |
+
Reward scaling: 1.0
|
| 15 |
+
Advantage temperature: 1.0
|
| 16 |
+
Use reference model: False
|
| 17 |
+
===================================
|
| 18 |
+
|
| 19 |
+
Starting GRPO training...
|
| 20 |
+
0%| | 0/500 [00:00<?, ?it/s]/workspace/trainer-kit/GRPO-14b/run_grpo_fixed.py:700: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
|
| 21 |
+
with torch.cuda.amp.autocast(enabled=self.args.fp16 or self.args.bf16):
|
| 22 |
+
Could not estimate the number of tokens of the input, floating-point operations will not be computed
|
| 23 |
+
20%|█████████████▍ | 100/500 [28:13<1:54:10, 17.13s/it]/workspace/trainer-kit/GRPO-14b/run_grpo_fixed.py:700: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
|
| 24 |
+
{'loss': 0.007, 'grad_norm': 0.05460292845964432, 'learning_rate': 8.000000000000001e-07, 'grpo_mean_advantage': -1.3560057254835556e-07, 'grpo_std_advantage': 3.0318567496578908e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5922331809997559, 'epoch': 0.02}
|
| 25 |
+
{'loss': 0.0107, 'grad_norm': 0.0679207444190979, 'learning_rate': 1.8000000000000001e-06, 'grpo_mean_advantage': 3.6619603633880615e-06, 'grpo_std_advantage': 1.6246918676188216e-05, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5561589002609253, 'epoch': 0.04}
|
| 26 |
+
{'loss': 0.007, 'grad_norm': 0.05788416787981987, 'learning_rate': 2.8000000000000003e-06, 'grpo_mean_advantage': -1.0654330395709621e-07, 'grpo_std_advantage': 5.399440965447866e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5759152173995972, 'epoch': 0.06}
|
| 27 |
+
{'loss': 0.0246, 'grad_norm': 0.0746568813920021, 'learning_rate': 3.8000000000000005e-06, 'grpo_mean_advantage': -5.871057737749652e-07, 'grpo_std_advantage': 2.6951597646984737e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5127314329147339, 'epoch': 0.08}
|
| 28 |
+
{'loss': 0.0337, 'grad_norm': 0.11442846059799194, 'learning_rate': 4.800000000000001e-06, 'grpo_mean_advantage': 6.370246410369873e-07, 'grpo_std_advantage': 2.8908377771585947e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.539706826210022, 'epoch': 0.1}
|
| 29 |
+
{'loss': 0.0171, 'grad_norm': 0.05778791010379791, 'learning_rate': 4.999125183044924e-06, 'grpo_mean_advantage': 6.705522359595761e-09, 'grpo_std_advantage': 6.189450800775376e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5812538862228394, 'epoch': 0.12}
|
| 30 |
+
{'loss': 0.0145, 'grad_norm': 0.05819695070385933, 'learning_rate': 4.995572288443412e-06, 'grpo_mean_advantage': 3.859400692363124e-07, 'grpo_std_advantage': 1.6833292875162442e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5909844636917114, 'epoch': 0.14}
|
| 31 |
+
{'loss': 0.0196, 'grad_norm': 0.07968433201313019, 'learning_rate': 4.98929052218411e-06, 'grpo_mean_advantage': 2.600252742013254e-07, 'grpo_std_advantage': 1.4095899132371414e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5630953907966614, 'epoch': 0.16}
|
| 32 |
+
{'loss': 0.0186, 'grad_norm': 0.0733402892947197, 'learning_rate': 4.980286753286196e-06, 'grpo_mean_advantage': -1.2591480924584175e-07, 'grpo_std_advantage': 1.0309080380466185e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5604403614997864, 'epoch': 0.18}
|
| 33 |
+
{'loss': 0.0286, 'grad_norm': 0.07136482000350952, 'learning_rate': 4.9685708272387645e-06, 'grpo_mean_advantage': -2.808868941883702e-07, 'grpo_std_advantage': 1.5696078889959608e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5971035957336426, 'epoch': 0.2}
|
| 34 |
+
{'loss': 0.0054, 'grad_norm': 0.08851475268602371, 'learning_rate': 4.9541555552349404e-06, 'grpo_mean_advantage': 2.6822089438383045e-08, 'grpo_std_advantage': 3.7878271541558206e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5892971754074097, 'epoch': 0.22}
|
| 35 |
+
{'loss': -0.0074, 'grad_norm': 0.07778509706258774, 'learning_rate': 4.9370567001630155e-06, 'grpo_mean_advantage': -5.662441182607836e-08, 'grpo_std_advantage': 6.128998393251095e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.564322292804718, 'epoch': 0.24}
|
| 36 |
+
{'loss': 0.0145, 'grad_norm': 0.08740051090717316, 'learning_rate': 4.917292959369968e-06, 'grpo_mean_advantage': -1.5944242193199898e-07, 'grpo_std_advantage': 1.6374274309782777e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.562497615814209, 'epoch': 0.26}
|
| 37 |
+
{'loss': 0.0257, 'grad_norm': 0.19070060551166534, 'learning_rate': 4.8948859442161876e-06, 'grpo_mean_advantage': 1.6838312433264946e-07, 'grpo_std_advantage': 8.536571272088622e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5904761552810669, 'epoch': 0.28}
|
| 38 |
+
{'loss': 0.0024, 'grad_norm': 0.07321271300315857, 'learning_rate': 4.869860156443768e-06, 'grpo_mean_advantage': 1.1175870895385742e-07, 'grpo_std_advantage': 6.451961667153228e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5765624046325684, 'epoch': 0.3}
|
| 39 |
+
{'loss': 0.0277, 'grad_norm': 0.07126748561859131, 'learning_rate': 4.842242961384211e-06, 'grpo_mean_advantage': -1.4603138254187797e-07, 'grpo_std_advantage': 1.1309343790344428e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5858271718025208, 'epoch': 0.32}
|
| 40 |
+
{'loss': 0.0246, 'grad_norm': 0.08629189431667328, 'learning_rate': 4.812064558034847e-06, 'grpo_mean_advantage': -1.817941665649414e-06, 'grpo_std_advantage': 1.1141768482048064e-05, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5871662497520447, 'epoch': 0.34}
|
| 41 |
+
{'loss': 0.0056, 'grad_norm': 0.0998779758810997, 'learning_rate': 4.779357946036662e-06, 'grpo_mean_advantage': 1.8179416372277046e-07, 'grpo_std_advantage': 6.210335072864837e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5330992937088013, 'epoch': 0.36}
|
| 42 |
+
{'loss': 0.0053, 'grad_norm': 0.10614689439535141, 'learning_rate': 4.74415888958968e-06, 'grpo_mean_advantage': -2.972781771859445e-07, 'grpo_std_advantage': 3.1582342217006953e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5265295505523682, 'epoch': 0.38}
|
| 43 |
+
{'loss': 0.0134, 'grad_norm': 0.10345634073019028, 'learning_rate': 4.706505878345343e-06, 'grpo_mean_advantage': -7.033348197182931e-07, 'grpo_std_advantage': 4.245831405569334e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5660771131515503, 'epoch': 0.4}
|
| 44 |
+
with torch.cuda.amp.autocast(enabled=self.args.fp16 or self.args.bf16):
|
| 45 |
+
40%|██████████████████████████▊ | 200/500 [56:13<1:25:51, 17.17s/it]/workspace/trainer-kit/GRPO-14b/run_grpo_fixed.py:700: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
|
| 46 |
+
{'loss': 0.0004, 'grad_norm': 0.10077933222055435, 'learning_rate': 4.666440085318626e-06, 'grpo_mean_advantage': 1.1920928955078125e-07, 'grpo_std_advantage': 3.2809634831210133e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.57631915807724, 'epoch': 0.42}
|
| 47 |
+
{'loss': 0.0033, 'grad_norm': 0.09548182785511017, 'learning_rate': 4.624005321865968e-06, 'grpo_mean_advantage': -4.0978193283081055e-07, 'grpo_std_advantage': 6.0397578636184335e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.546563982963562, 'epoch': 0.44}
|
| 48 |
+
{'loss': 0.0095, 'grad_norm': 0.09417816251516342, 'learning_rate': 4.57924798977818e-06, 'grpo_mean_advantage': -1.467764434437413e-07, 'grpo_std_advantage': 2.2689375782647403e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5519219636917114, 'epoch': 0.46}
|
| 49 |
+
{'loss': 0.0006, 'grad_norm': 0.10022275149822235, 'learning_rate': 4.532217030540781e-06, 'grpo_mean_advantage': -5.215406329028838e-09, 'grpo_std_advantage': 7.929010621410271e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5490407943725586, 'epoch': 0.48}
|
| 50 |
+
{'loss': -0.0046, 'grad_norm': 0.14057794213294983, 'learning_rate': 4.482963871817195e-06, 'grpo_mean_advantage': -5.7369469175228005e-08, 'grpo_std_advantage': 1.2823379620385822e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5646580457687378, 'epoch': 0.5}
|
| 51 |
+
{'loss': -0.003, 'grad_norm': 0.12420658767223358, 'learning_rate': 4.4315423712133595e-06, 'grpo_mean_advantage': 2.9876827056796174e-07, 'grpo_std_advantage': 1.0496698905626545e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.6111599802970886, 'epoch': 0.52}
|
| 52 |
+
{'loss': 0.0154, 'grad_norm': 0.14342808723449707, 'learning_rate': 4.378008757385222e-06, 'grpo_mean_advantage': 1.5869736103013565e-07, 'grpo_std_advantage': 1.2748531617035042e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5619662404060364, 'epoch': 0.54}
|
| 53 |
+
{'loss': -0.0262, 'grad_norm': 0.14729444682598114, 'learning_rate': 4.322421568553529e-06, 'grpo_mean_advantage': 3.0100346748440643e-07, 'grpo_std_advantage': 2.4499684059264837e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5795454978942871, 'epoch': 0.56}
|
| 54 |
+
{'loss': 0.0018, 'grad_norm': 0.15249410271644592, 'learning_rate': 4.2648415884931476e-06, 'grpo_mean_advantage': -3.233552092751779e-07, 'grpo_std_advantage': 1.248456669600273e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5804953575134277, 'epoch': 0.58}
|
| 55 |
+
{'loss': -0.017, 'grad_norm': 0.1841023564338684, 'learning_rate': 4.205331780066892e-06, 'grpo_mean_advantage': 3.2261013416245987e-07, 'grpo_std_advantage': 1.4773489738217904e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5628539323806763, 'epoch': 0.6}
|
| 56 |
+
{'loss': 0.0044, 'grad_norm': 0.18597163259983063, 'learning_rate': 4.1439572163765615e-06, 'grpo_mean_advantage': -2.5331974029541016e-07, 'grpo_std_advantage': 1.5092309695319273e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5727725625038147, 'epoch': 0.62}
|
| 57 |
+
{'loss': -0.005, 'grad_norm': 0.18310388922691345, 'learning_rate': 4.0807850096064605e-06, 'grpo_mean_advantage': -6.780028627417778e-08, 'grpo_std_advantage': 8.550978805033083e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5833909511566162, 'epoch': 0.64}
|
| 58 |
+
{'loss': -0.015, 'grad_norm': 0.2192923128604889, 'learning_rate': 4.015884237637206e-06, 'grpo_mean_advantage': -5.587935447692871e-08, 'grpo_std_advantage': 3.564579174053506e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5742615461349487, 'epoch': 0.66}
|
| 59 |
+
{'loss': -0.0314, 'grad_norm': 0.16708803176879883, 'learning_rate': 3.949325868510083e-06, 'grpo_mean_advantage': -5.327165126800537e-07, 'grpo_std_advantage': 2.309018327650847e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5758188962936401, 'epoch': 0.68}
|
| 60 |
+
{'loss': -0.0441, 'grad_norm': 0.3401262164115906, 'learning_rate': 3.881182682824534e-06, 'grpo_mean_advantage': 5.863606702405377e-07, 'grpo_std_advantage': 2.4449204829579685e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5767683982849121, 'epoch': 0.7}
|
| 61 |
+
{'loss': -0.0162, 'grad_norm': 0.1931898146867752, 'learning_rate': 3.811529194153635e-06, 'grpo_mean_advantage': 3.2186508747145126e-07, 'grpo_std_advantage': 2.293551688126172e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.586772084236145, 'epoch': 0.72}
|
| 62 |
+
{'loss': -0.0386, 'grad_norm': 0.2537969648838043, 'learning_rate': 3.7404415675646054e-06, 'grpo_mean_advantage': -4.470348358154297e-08, 'grpo_std_advantage': 3.7067667335577426e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.549396276473999, 'epoch': 0.74}
|
| 63 |
+
{'loss': -0.037, 'grad_norm': 0.20326584577560425, 'learning_rate': 3.667997536333424e-06, 'grpo_mean_advantage': -2.1010637851759384e-07, 'grpo_std_advantage': 1.1695076409523608e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5798425078392029, 'epoch': 0.76}
|
| 64 |
+
{'loss': -0.0292, 'grad_norm': 0.25048357248306274, 'learning_rate': 3.59427631694463e-06, 'grpo_mean_advantage': 1.765787658314366e-07, 'grpo_std_advantage': 2.429934738756856e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5584167838096619, 'epoch': 0.78}
|
| 65 |
+
{'loss': -0.0454, 'grad_norm': 0.2687569260597229, 'learning_rate': 3.5193585224692595e-06, 'grpo_mean_advantage': 1.6540289493605087e-07, 'grpo_std_advantage': 2.6342788714828203e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5676193237304688, 'epoch': 0.8}
|
| 66 |
+
with torch.cuda.amp.autocast(enabled=self.args.fp16 or self.args.bf16):
|
| 67 |
+
60%|████████████████████████████████████████▏ | 300/500 [1:24:01<55:53, 16.77s/it]/workspace/trainer-kit/GRPO-14b/run_grpo_fixed.py:700: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
|
| 68 |
+
{'loss': -0.0423, 'grad_norm': 0.22301620244979858, 'learning_rate': 3.44332607441564e-06, 'grpo_mean_advantage': -1.0944902442133753e-06, 'grpo_std_advantage': 5.346942998585291e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5669739842414856, 'epoch': 0.82}
|
| 69 |
+
{'loss': -0.0857, 'grad_norm': 0.3040211498737335, 'learning_rate': 3.3662621131494204e-06, 'grpo_mean_advantage': 2.4065374759629776e-07, 'grpo_std_advantage': 1.6327536513927043e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5922158360481262, 'epoch': 0.84}
|
| 70 |
+
{'loss': -0.0278, 'grad_norm': 0.27231141924858093, 'learning_rate': 3.2882509069808044e-06, 'grpo_mean_advantage': -5.21540641784668e-08, 'grpo_std_advantage': 5.847922466273303e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5473950505256653, 'epoch': 0.86}
|
| 71 |
+
{'loss': -0.0727, 'grad_norm': 0.3571636378765106, 'learning_rate': 3.2093777600183873e-06, 'grpo_mean_advantage': 6.541609991472797e-07, 'grpo_std_advantage': 4.072162937518442e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5880032777786255, 'epoch': 0.88}
|
| 72 |
+
{'loss': -0.0464, 'grad_norm': 0.306273490190506, 'learning_rate': 3.1297289188903705e-06, 'grpo_mean_advantage': -1.2218951894737984e-07, 'grpo_std_advantage': 4.386006935419573e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5835092663764954, 'epoch': 0.9}
|
| 73 |
+
{'loss': -0.0295, 'grad_norm': 0.2700377106666565, 'learning_rate': 3.049391478435133e-06, 'grpo_mean_advantage': 1.7605722177904681e-06, 'grpo_std_advantage': 8.007580618141219e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5394966006278992, 'epoch': 0.92}
|
| 74 |
+
{'loss': -0.031, 'grad_norm': 0.39531761407852173, 'learning_rate': 2.9684532864643123e-06, 'grpo_mean_advantage': -3.3080578987210174e-07, 'grpo_std_advantage': 1.551636614749441e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5687432289123535, 'epoch': 0.94}
|
| 75 |
+
{'loss': -0.0789, 'grad_norm': 0.5987040996551514, 'learning_rate': 2.887002847702504e-06, 'grpo_mean_advantage': 2.712011450967111e-07, 'grpo_std_advantage': 1.4400844747797237e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5550583600997925, 'epoch': 0.96}
|
| 76 |
+
{'loss': -0.1131, 'grad_norm': 0.5680716037750244, 'learning_rate': 2.8051292270086506e-06, 'grpo_mean_advantage': -3.2857059295565705e-07, 'grpo_std_advantage': 2.105091425619321e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.558111310005188, 'epoch': 0.98}
|
| 77 |
+
{'loss': -0.2232, 'grad_norm': 0.6204046010971069, 'learning_rate': 2.722921951984927e-06, 'grpo_mean_advantage': 4.470348358154297e-08, 'grpo_std_advantage': 5.315724820320611e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.6196198463439941, 'epoch': 1.0}
|
| 78 |
+
{'loss': -0.1363, 'grad_norm': 0.8389026522636414, 'learning_rate': 2.640470915079614e-06, 'grpo_mean_advantage': 9.290873776990338e-07, 'grpo_std_advantage': 4.219644324621186e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.582168459892273, 'epoch': 1.02}
|
| 79 |
+
{'loss': -0.1868, 'grad_norm': 0.9067686796188354, 'learning_rate': 2.557866275291035e-06, 'grpo_mean_advantage': 2.533197474008375e-08, 'grpo_std_advantage': 1.6600588992332632e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5551307797431946, 'epoch': 1.04}
|
| 80 |
+
{'loss': -0.1792, 'grad_norm': 0.9277902841567993, 'learning_rate': 2.4751983595800093e-06, 'grpo_mean_advantage': -5.662441182607836e-08, 'grpo_std_advantage': 1.0909400316450046e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.535040020942688, 'epoch': 1.06}
|
| 81 |
+
{'loss': -0.1691, 'grad_norm': 1.0715463161468506, 'learning_rate': 2.392557564098649e-06, 'grpo_mean_advantage': -9.536743306171047e-08, 'grpo_std_advantage': 5.838213610331877e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5673571825027466, 'epoch': 1.08}
|
| 82 |
+
{'loss': -0.1655, 'grad_norm': 0.7759184837341309, 'learning_rate': 2.3100342553434924e-06, 'grpo_mean_advantage': 3.278255533700758e-08, 'grpo_std_advantage': 9.317170679423725e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5874732732772827, 'epoch': 1.1}
|
| 83 |
+
{'loss': -0.1821, 'grad_norm': 0.9387398958206177, 'learning_rate': 2.2277186713410688e-06, 'grpo_mean_advantage': -1.206994113545079e-07, 'grpo_std_advantage': 6.201085511747806e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5569106340408325, 'epoch': 1.12}
|
| 84 |
+
{'loss': -0.2102, 'grad_norm': 1.6132302284240723, 'learning_rate': 2.1457008229739395e-06, 'grpo_mean_advantage': 4.470348358154297e-08, 'grpo_std_advantage': 6.115651558502577e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5578873157501221, 'epoch': 1.14}
|
| 85 |
+
{'loss': -0.2937, 'grad_norm': 0.8679026961326599, 'learning_rate': 2.0640703955551214e-06, 'grpo_mean_advantage': -3.3453108017056365e-07, 'grpo_std_advantage': 3.5326345368957845e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5735999345779419, 'epoch': 1.16}
|
| 86 |
+
{'loss': -0.2598, 'grad_norm': 1.0550166368484497, 'learning_rate': 1.9829166507585084e-06, 'grpo_mean_advantage': -1.110136480519941e-07, 'grpo_std_advantage': 4.731904823529476e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5626259446144104, 'epoch': 1.18}
|
| 87 |
+
{'loss': -0.2546, 'grad_norm': 1.2819372415542603, 'learning_rate': 1.90232832901255e-06, 'grpo_mean_advantage': -5.08874677507265e-07, 'grpo_std_advantage': 1.840126174101897e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5463050603866577, 'epoch': 1.2}
|
| 88 |
+
with torch.cuda.amp.autocast(enabled=self.args.fp16 or self.args.bf16):
|
| 89 |
+
80%|█████████████████████████████████████████████████████▌ | 400/500 [1:52:15<28:44, 17.25s/it]/workspace/trainer-kit/GRPO-14b/run_grpo_fixed.py:700: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
|
| 90 |
+
{'loss': -0.1809, 'grad_norm': 1.0188143253326416, 'learning_rate': 1.82239355246389e-06, 'grpo_mean_advantage': 1.01327898960335e-07, 'grpo_std_advantage': 7.798533943059738e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5352144241333008, 'epoch': 1.22}
|
| 91 |
+
{'loss': -0.3559, 'grad_norm': 2.0709052085876465, 'learning_rate': 1.7431997286170923e-06, 'grpo_mean_advantage': 1.341104507446289e-07, 'grpo_std_advantage': 7.821902840987605e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5547868013381958, 'epoch': 1.24}
|
| 92 |
+
{'loss': -0.3874, 'grad_norm': 1.8516215085983276, 'learning_rate': 1.6648334547558227e-06, 'grpo_mean_advantage': 9.015202806494926e-08, 'grpo_std_advantage': 1.0693488547985908e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5859472751617432, 'epoch': 1.26}
|
| 93 |
+
{'loss': -0.3467, 'grad_norm': 1.283104419708252, 'learning_rate': 1.5873804232499862e-06, 'grpo_mean_advantage': -2.443790378947597e-07, 'grpo_std_advantage': 1.183122208203713e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5751550793647766, 'epoch': 1.28}
|
| 94 |
+
{'loss': -0.1703, 'grad_norm': 1.4108576774597168, 'learning_rate': 1.51092532785238e-06, 'grpo_mean_advantage': -6.705522537231445e-08, 'grpo_std_advantage': 6.109748937888071e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5497723817825317, 'epoch': 1.3}
|
| 95 |
+
{'loss': -0.2918, 'grad_norm': 1.0421361923217773, 'learning_rate': 1.4355517710873184e-06, 'grpo_mean_advantage': -1.639127766850379e-08, 'grpo_std_advantage': 5.529495297196263e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.55989670753479, 'epoch': 1.32}
|
| 96 |
+
{'loss': -0.3069, 'grad_norm': 1.3465828895568848, 'learning_rate': 1.361342172832502e-06, 'grpo_mean_advantage': 4.418194237132411e-07, 'grpo_std_advantage': 2.9275292945385445e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5809233784675598, 'epoch': 1.34}
|
| 97 |
+
{'loss': -0.5594, 'grad_norm': 1.1959459781646729, 'learning_rate': 1.2883776801940884e-06, 'grpo_mean_advantage': 9.685754776000977e-08, 'grpo_std_advantage': 3.754235251562932e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5568087100982666, 'epoch': 1.36}
|
| 98 |
+
{'loss': -0.4102, 'grad_norm': 1.8967422246932983, 'learning_rate': 1.216738078773522e-06, 'grpo_mean_advantage': -2.384185791015625e-07, 'grpo_std_advantage': 6.821086913078034e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5655568838119507, 'epoch': 1.38}
|
| 99 |
+
{'loss': -0.338, 'grad_norm': 2.221132755279541, 'learning_rate': 1.146501705423155e-06, 'grpo_mean_advantage': -8.717179156292332e-08, 'grpo_std_advantage': 2.500940354366321e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.6089578866958618, 'epoch': 1.4}
|
| 100 |
+
{'loss': -0.4985, 'grad_norm': 2.3640377521514893, 'learning_rate': 1.0777453625860474e-06, 'grpo_mean_advantage': 2.1606683731079102e-07, 'grpo_std_advantage': 1.4568390724889468e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.6129671335220337, 'epoch': 1.42}
|
| 101 |
+
{'loss': -0.4347, 'grad_norm': 1.9084734916687012, 'learning_rate': 1.0105442343136184e-06, 'grpo_mean_advantage': -3.725290298461914e-09, 'grpo_std_advantage': 2.965894054796081e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5562310814857483, 'epoch': 1.44}
|
| 102 |
+
{'loss': -0.6217, 'grad_norm': 1.6063904762268066, 'learning_rate': 9.449718040529987e-07, 'grpo_mean_advantage': 4.313886279305734e-07, 'grpo_std_advantage': 1.9621948013082147e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5884170532226562, 'epoch': 1.46}
|
| 103 |
+
{'loss': -0.5364, 'grad_norm': 2.114664077758789, 'learning_rate': 8.810997742939531e-07, 'grpo_mean_advantage': 2.0489096641540527e-07, 'grpo_std_advantage': 1.0235522722723545e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5795440673828125, 'epoch': 1.48}
|
| 104 |
+
{'loss': -0.4798, 'grad_norm': 1.8450465202331543, 'learning_rate': 8.189979881632634e-07, 'grpo_mean_advantage': -1.4185905001795618e-06, 'grpo_std_advantage': 1.0947338523692451e-05, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5607603788375854, 'epoch': 1.5}
|
| 105 |
+
{'loss': -0.4805, 'grad_norm': 2.673438787460327, 'learning_rate': 7.587343530522945e-07, 'grpo_mean_advantage': -1.758337049295733e-07, 'grpo_std_advantage': 9.663675655247062e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5381432771682739, 'epoch': 1.52}
|
| 106 |
+
{'loss': -0.433, 'grad_norm': 2.2263550758361816, 'learning_rate': 7.003747663612581e-07, 'grpo_mean_advantage': -6.973743325033865e-07, 'grpo_std_advantage': 4.341973180999048e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5528443455696106, 'epoch': 1.54}
|
| 107 |
+
{'loss': -0.6021, 'grad_norm': 2.3657093048095703, 'learning_rate': 6.439830434413754e-07, 'grpo_mean_advantage': 1.7881394143159923e-08, 'grpo_std_advantage': 1.3004198251564958e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.6091476678848267, 'epoch': 1.56}
|
| 108 |
+
{'loss': -0.5595, 'grad_norm': 1.9847129583358765, 'learning_rate': 5.896208478137222e-07, 'grpo_mean_advantage': 3.4868716625169327e-07, 'grpo_std_advantage': 2.059372718576924e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5397372245788574, 'epoch': 1.58}
|
| 109 |
+
{'loss': -0.5592, 'grad_norm': 2.922114133834839, 'learning_rate': 5.373476237410808e-07, 'grpo_mean_advantage': -2.1636485598719446e-06, 'grpo_std_advantage': 9.725940799398813e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5873125195503235, 'epoch': 1.6}
|
| 110 |
+
with torch.cuda.amp.autocast(enabled=self.args.fp16 or self.args.bf16):
|
| 111 |
+
100%|███████████████████████████████████████████████████████████████████| 500/500 [2:20:30<00:00, 16.86s/it]
|
| 112 |
+
{'loss': -0.5623, 'grad_norm': 1.8524045944213867, 'learning_rate': 4.872205312265074e-07, 'grpo_mean_advantage': -5.960464477539063e-08, 'grpo_std_advantage': 3.460792754594877e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5601426362991333, 'epoch': 1.62}
|
| 113 |
+
{'loss': -0.5943, 'grad_norm': 1.7269790172576904, 'learning_rate': 4.3929438350970687e-07, 'grpo_mean_advantage': 2.6226044269606064e-07, 'grpo_std_advantage': 7.928817922220333e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.578656792640686, 'epoch': 1.64}
|
| 114 |
+
{'loss': -0.6193, 'grad_norm': 2.26530122756958, 'learning_rate': 3.936215871295634e-07, 'grpo_mean_advantage': 2.3558736756967846e-06, 'grpo_std_advantage': 1.4469559573626611e-05, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5885810852050781, 'epoch': 1.66}
|
| 115 |
+
{'loss': -0.6934, 'grad_norm': 2.6794464588165283, 'learning_rate': 3.502520846183577e-07, 'grpo_mean_advantage': 1.639127766850379e-08, 'grpo_std_advantage': 9.352411325380672e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5805023312568665, 'epoch': 1.68}
|
| 116 |
+
{'loss': -0.5126, 'grad_norm': 2.100447654724121, 'learning_rate': 3.092332998903416e-07, 'grpo_mean_advantage': 3.2387674764322583e-06, 'grpo_std_advantage': 1.999079904635437e-05, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5655918121337891, 'epoch': 1.7}
|
| 117 |
+
{'loss': -0.5446, 'grad_norm': 2.1027915477752686, 'learning_rate': 2.706100863843822e-07, 'grpo_mean_advantage': 3.5464762504489045e-07, 'grpo_std_advantage': 1.7663603557593888e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5474504232406616, 'epoch': 1.72}
|
| 118 |
+
{'loss': -0.5125, 'grad_norm': 2.289045572280884, 'learning_rate': 2.3442467801738867e-07, 'grpo_mean_advantage': 3.6135315895080566e-07, 'grpo_std_advantage': 2.356920958845876e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5874254703521729, 'epoch': 1.74}
|
| 119 |
+
{'loss': -0.595, 'grad_norm': 2.278038501739502, 'learning_rate': 2.007166430021415e-07, 'grpo_mean_advantage': 2.7567148563889532e-08, 'grpo_std_advantage': 9.97340521280421e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5815118551254272, 'epoch': 1.76}
|
| 120 |
+
{'loss': -0.8055, 'grad_norm': 2.340942621231079, 'learning_rate': 1.6952284058003366e-07, 'grpo_mean_advantage': -8.34465012644614e-08, 'grpo_std_advantage': 5.558832185670326e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5611211061477661, 'epoch': 1.78}
|
| 121 |
+
{'loss': -0.8561, 'grad_norm': 2.4256298542022705, 'learning_rate': 1.4087738071603075e-07, 'grpo_mean_advantage': -1.9818544672034477e-07, 'grpo_std_advantage': 6.800727305744658e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.590424656867981, 'epoch': 1.8}
|
| 122 |
+
{'loss': -0.429, 'grad_norm': 1.6453255414962769, 'learning_rate': 1.1481158679992554e-07, 'grpo_mean_advantage': -1.9371508841459217e-08, 'grpo_std_advantage': 3.142378943721269e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5670351982116699, 'epoch': 1.82}
|
| 123 |
+
{'loss': -0.642, 'grad_norm': 2.3458049297332764, 'learning_rate': 9.135396139467151e-08, 'grpo_mean_advantage': 2.3692845729783585e-07, 'grpo_std_advantage': 1.682946731307311e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5640432834625244, 'epoch': 1.84}
|
| 124 |
+
{'loss': -0.583, 'grad_norm': 2.730945110321045, 'learning_rate': 7.053015506924749e-08, 'grpo_mean_advantage': 1.110136480519941e-07, 'grpo_std_advantage': 8.930008448260196e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5584251284599304, 'epoch': 1.86}
|
| 125 |
+
{'loss': -0.5197, 'grad_norm': 2.1463465690612793, 'learning_rate': 5.236293835013839e-08, 'grpo_mean_advantage': 2.5406478698641877e-07, 'grpo_std_advantage': 9.93092498902115e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5392154455184937, 'epoch': 1.88}
|
| 126 |
+
{'loss': -0.5864, 'grad_norm': 2.427900791168213, 'learning_rate': 3.687217682209837e-08, 'grpo_mean_advantage': -8.940697071579962e-09, 'grpo_std_advantage': 5.835169645251881e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5686308741569519, 'epoch': 1.9}
|
| 127 |
+
{'loss': -0.5721, 'grad_norm': 2.042795419692993, 'learning_rate': 2.4074809405425227e-08, 'grpo_mean_advantage': 4.0605664253234863e-07, 'grpo_std_advantage': 2.3210795916384086e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5842767357826233, 'epoch': 1.92}
|
| 128 |
+
{'loss': -0.5944, 'grad_norm': 2.800136089324951, 'learning_rate': 1.3984829833499636e-08, 'grpo_mean_advantage': 1.341104507446289e-07, 'grpo_std_advantage': 1.507950400991831e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5495311617851257, 'epoch': 1.94}
|
| 129 |
+
{'loss': -0.7015, 'grad_norm': 2.8475866317749023, 'learning_rate': 6.6132713508446075e-09, 'grpo_mean_advantage': 2.689659481802664e-07, 'grpo_std_advantage': 8.491958851664094e-07, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.549436628818512, 'epoch': 1.96}
|
| 130 |
+
{'loss': -0.4033, 'grad_norm': 2.9422402381896973, 'learning_rate': 1.9681946484320645e-09, 'grpo_mean_advantage': 8.195638656616211e-08, 'grpo_std_advantage': 3.802849732892355e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.544632077217102, 'epoch': 1.98}
|
| 131 |
+
{'loss': -0.6773, 'grad_norm': 2.66204833984375, 'learning_rate': 5.467904943851077e-11, 'grpo_mean_advantage': 7.552536089860951e-07, 'grpo_std_advantage': 4.143997102801222e-06, 'grpo_mean_kl_div': 0.0, 'grpo_mean_group_score': 0.5968535542488098, 'epoch': 2.0}
|
| 132 |
+
{'train_runtime': 8430.3461, 'train_samples_per_second': 0.474, 'train_steps_per_second': 0.059, 'train_loss': -0.23323759501613678, 'epoch': 2.0}
|
| 133 |
+
Saved best adapter -> runs/grpo_14b_run1/best_adapter
|
| 134 |
+
Merge disabled. Run with --merge-only later if needed.
|
grpo_qwen_14B_v2/wandb/run-20251227_194423-jz7bptqa/files/requirements.txt
ADDED
|
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
exceptiongroup==1.3.1
|
| 2 |
+
wheel==0.45.1
|
| 3 |
+
python-dateutil==2.9.0.post0
|
| 4 |
+
huggingface-hub==0.36.0
|
| 5 |
+
idna==3.11
|
| 6 |
+
click==8.3.1
|
| 7 |
+
numpy==2.2.6
|
| 8 |
+
httpx==0.28.1
|
| 9 |
+
tokenizers==0.22.1
|
| 10 |
+
sympy==1.13.1
|
| 11 |
+
yarl==1.22.0
|
| 12 |
+
async-timeout==5.0.1
|
| 13 |
+
datasets==4.4.2
|
| 14 |
+
einops==0.8.1
|
| 15 |
+
platformdirs==4.5.1
|
| 16 |
+
nvidia-cuda-cupti-cu12==12.1.105
|
| 17 |
+
nvidia-nvtx-cu12==12.1.105
|
| 18 |
+
smmap==5.0.2
|
| 19 |
+
accelerate==1.12.0
|
| 20 |
+
requests==2.32.5
|
| 21 |
+
aiohttp==3.13.2
|
| 22 |
+
transformers==4.57.3
|
| 23 |
+
bitsandbytes==0.49.0
|
| 24 |
+
evaluate==0.4.6
|
| 25 |
+
nvidia-cublas-cu12==12.1.3.1
|
| 26 |
+
mpmath==1.3.0
|
| 27 |
+
torchaudio==2.5.1+cu121
|
| 28 |
+
nvidia-cuda-runtime-cu12==12.1.105
|
| 29 |
+
typing-inspection==0.4.2
|
| 30 |
+
GitPython==3.1.45
|
| 31 |
+
xxhash==3.6.0
|
| 32 |
+
nvidia-cusolver-cu12==11.4.5.107
|
| 33 |
+
pydantic_core==2.41.5
|
| 34 |
+
six==1.17.0
|
| 35 |
+
torchvision==0.20.1+cu121
|
| 36 |
+
typing_extensions==4.15.0
|
| 37 |
+
triton==3.1.0
|
| 38 |
+
charset-normalizer==3.4.4
|
| 39 |
+
wandb==0.23.1
|
| 40 |
+
regex==2025.11.3
|
| 41 |
+
pip==25.3
|
| 42 |
+
nvidia-cusparse-cu12==12.1.0.106
|
| 43 |
+
pytz==2025.2
|
| 44 |
+
Jinja2==3.1.6
|
| 45 |
+
psutil==7.2.0
|
| 46 |
+
pillow==12.0.0
|
| 47 |
+
packaging==25.0
|
| 48 |
+
safetensors==0.7.0
|
| 49 |
+
sentry-sdk==2.48.0
|
| 50 |
+
gitdb==4.0.12
|
| 51 |
+
httpcore==1.0.9
|
| 52 |
+
setuptools==80.9.0
|
| 53 |
+
nvidia-cufft-cu12==11.0.2.54
|
| 54 |
+
flash_attn==2.8.3
|
| 55 |
+
anyio==4.12.0
|
| 56 |
+
pydantic==2.12.5
|
| 57 |
+
fsspec==2025.10.0
|
| 58 |
+
filelock==3.20.0
|
| 59 |
+
PyYAML==6.0.3
|
| 60 |
+
hf-xet==1.2.0
|
| 61 |
+
nvidia-cudnn-cu12==9.1.0.70
|
| 62 |
+
tqdm==4.67.1
|
| 63 |
+
MarkupSafe==2.1.5
|
| 64 |
+
attrs==25.4.0
|
| 65 |
+
nvidia-cuda-nvrtc-cu12==12.1.105
|
| 66 |
+
peft==0.18.0
|
| 67 |
+
aiohappyeyeballs==2.6.1
|
| 68 |
+
networkx==3.4.2
|
| 69 |
+
nvidia-nvjitlink-cu12==12.9.86
|
| 70 |
+
certifi==2025.11.12
|
| 71 |
+
pyarrow==22.0.0
|
| 72 |
+
dill==0.4.0
|
| 73 |
+
protobuf==6.33.2
|
| 74 |
+
aiosignal==1.4.0
|
| 75 |
+
frozenlist==1.8.0
|
| 76 |
+
urllib3==2.6.2
|
| 77 |
+
propcache==0.4.1
|
| 78 |
+
tzdata==2025.3
|
| 79 |
+
pandas==2.3.3
|
| 80 |
+
annotated-types==0.7.0
|
| 81 |
+
nvidia-nccl-cu12==2.21.5
|
| 82 |
+
multidict==6.7.0
|
| 83 |
+
nvidia-curand-cu12==10.3.2.106
|
| 84 |
+
torch==2.5.1+cu121
|
| 85 |
+
h11==0.16.0
|
| 86 |
+
multiprocess==0.70.18
|
| 87 |
+
wheel==0.45.1
|
| 88 |
+
tomli==2.0.1
|
| 89 |
+
autocommand==2.2.2
|
| 90 |
+
jaraco.context==5.3.0
|
| 91 |
+
zipp==3.19.2
|
| 92 |
+
packaging==24.2
|
| 93 |
+
inflect==7.3.1
|
| 94 |
+
typing_extensions==4.12.2
|
| 95 |
+
platformdirs==4.2.2
|
| 96 |
+
jaraco.functools==4.0.1
|
| 97 |
+
jaraco.collections==5.1.0
|
| 98 |
+
jaraco.text==3.12.1
|
| 99 |
+
backports.tarfile==1.2.0
|
| 100 |
+
more-itertools==10.3.0
|
| 101 |
+
importlib_metadata==8.0.0
|
| 102 |
+
typeguard==4.3.0
|