Completed SFT training (5/5 epochs). Preparing for multi-round DPO training.

README.md

@@ -53,7 +53,7 @@ model-index:
     metrics:
     - name: single choice
       type: exact-match
-      value: 28.71
+      value: 31.1
 widget:
 - text: 中華民國憲法第一條
 metrics:
@@ -76,8 +76,9 @@ new_version: lianghsun/Llama-3.2-Taiwan-3B-Instruct
   
   | Update Date  | Model Version         | Key Changes                         |
   |--------------|-----------------------|-------------------------------------|
-  | 2024/11/25   | v2024.11.25           | Updated model version to v2024.11.25, training progressed to (3/10) epochs. Still in SFT stage, DPO training remains pending. |
-  | 2024/11/22   | v2024.11.22           |Initial upload: Model version v2024.11.22, training completed up to (1/10) epochs. Currently trained only on SFT, DPO training not yet performed. |
+  | 2024/11/27   | v2024.11.27           | Completed SFT training (5/5 epochs). Preparing for multi-round DPO training. |
+  | 2024/11/25   | v2024.11.25           | Updated model version to v2024.11.25, training progressed to (3/5) epochs. Still in SFT stage, DPO training remains pending. |
+  | 2024/11/22   | v2024.11.22           | Initial upload: Model version v2024.11.22, training completed up to (1/5) epochs. Currently trained only on SFT, DPO training not yet performed. |
   
 </details>
 
@@ -223,17 +224,17 @@ docker run --runtime nvidia --gpus all \
 - **lr_scheduler_type:** cosine
 - **lr_scheduler_warmup_ratio:** 0.01
 - **num_epochs:** 5.0
+- **global_step:** 590
 
 #### Speeds, Sizes, Times
 
 <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-*(WIP)*
-- **Duration**:
-- **Train runtime**:
-- **Train samples per second**: 
-- **Train steps per second**:
-- **Total training FLOPs**:
-- **Train loss**: 
+- **Duration**: 5 days, 16:15:11.17
+- **Train runtime**: 490,511.1789
+- **Train samples per second**: 25.37
+- **Train steps per second**: 0.001
+- **Total training FLOPs**: 26,658,386,120,540,160
+- **Train loss**: 0.8533
 
 ## Evaluation
 
@@ -347,4 +348,4 @@ base_model: lianghsun/Llama-3.2-Taiwan-3B-Instruct
 - Transformers 4.45.2
 - Pytorch 2.4.1+cu121
 - Datasets 2.21.0
-- Tokenizers 0.20.0
+- Tokenizers 0.20.0

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "/home/owen/LLaMA-Factory/saves/llama3.2-3B/sft/full/sft_2024-11-21-1/checkpoint-10",
+  "_name_or_path": "lianghsun/Llama-3.2-Taiwan-3B-Instruct",
   "architectures": [
     "LlamaForCausalLM"
   ],

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:969c04d243dc2a80b40c3edb3e0391567710eecce407b22006c73fadbe8fb455
+oid sha256:2a4129d85cadd57a2bd201e67d424bf472aad2d1f708fc479d6888ff859dca4d
 size 4965799096

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c0133ef2cccfbf9d713da104bd8a27882674ad1bde198ae7e4b5b4d005e85df
+oid sha256:7f4a01c12a64842a84ab1aea11fc242f3cd7f305931c5e86385ca18896ede088
 size 1459729952