Upload hyperparams.json with huggingface_hub
Browse files- hyperparams.json +35 -0
hyperparams.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dtype": "torch.float16",
|
| 3 |
+
"train_dataset": "EYLSFTStaticDataset",
|
| 4 |
+
"train_dataset_len": 9223372036854775807,
|
| 5 |
+
"test_dataset": "EYLSFTStaticDataset",
|
| 6 |
+
"test_dataset_len": 9223372036854775807,
|
| 7 |
+
"n_layers": "24",
|
| 8 |
+
"n_heads": "16",
|
| 9 |
+
"embedding_dim": "1024",
|
| 10 |
+
"dropout_rate": "0.2",
|
| 11 |
+
"use_bias": "True",
|
| 12 |
+
"block_size": "1024",
|
| 13 |
+
"vocab_size": "50257",
|
| 14 |
+
"model_name": "gpt2-medium/dropout",
|
| 15 |
+
"hf_model": "gpt2-medium",
|
| 16 |
+
"grad_clip": "1.0",
|
| 17 |
+
"exp_name": "experiment_name",
|
| 18 |
+
"batch_size": "4",
|
| 19 |
+
"lr": "0.0001",
|
| 20 |
+
"lora_rank": "0",
|
| 21 |
+
"pretrain": "huggingface",
|
| 22 |
+
"activation_checkpointing": "False",
|
| 23 |
+
"finetune_method": "",
|
| 24 |
+
"total_epochs": "1",
|
| 25 |
+
"max_steps": "50000",
|
| 26 |
+
"actor_weights": "",
|
| 27 |
+
"critic_weights": "",
|
| 28 |
+
"reward_model_weights": "",
|
| 29 |
+
"sft_model_weights": "",
|
| 30 |
+
"actor_lr": "5e-06",
|
| 31 |
+
"critic_lr": "9e-06",
|
| 32 |
+
"kl_beta": "0.02",
|
| 33 |
+
"adam_beta1": "0.9",
|
| 34 |
+
"adam_beta2": "0.95"
|
| 35 |
+
}
|