Model save

README.md

@@ -1,13 +1,11 @@
 ---
 base_model: deepseek-ai/Deepseek-R1-Distill-Qwen-7B
-datasets:
-- tttx/r1-trajectories-collection-round-2
 library_name: peft
 license: mit
 tags:
-- alignment-handbook
 - trl
 - sft
+- alignment-handbook
 - generated_from_trainer
 model-index:
 - name: sft_r1_7b
@@ -19,9 +17,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # sft_r1_7b
 
-This model is a fine-tuned version of [deepseek-ai/Deepseek-R1-Distill-Qwen-7B](https://huggingface.co/deepseek-ai/Deepseek-R1-Distill-Qwen-7B) on the tttx/r1-trajectories-collection-round-2 dataset.
-It achieves the following results on the evaluation set:
-- Loss: 0.4983
+This model is a fine-tuned version of [deepseek-ai/Deepseek-R1-Distill-Qwen-7B](https://huggingface.co/deepseek-ai/Deepseek-R1-Distill-Qwen-7B) on the None dataset.
 
 ## Model description
 
@@ -42,23 +38,19 @@ More information needed
 The following hyperparameters were used during training:
 - learning_rate: 5e-05
 - train_batch_size: 8
-- eval_batch_size: 1
+- eval_batch_size: 8
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 4
 - total_train_batch_size: 32
-- total_eval_batch_size: 4
+- total_eval_batch_size: 32
 - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 2
+- num_epochs: 3
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss |
-|:-------------:|:-----:|:----:|:---------------:|
-| 0.4731        | 1.0   | 92   | 0.5096          |
-| 0.4312        | 2.0   | 184  | 0.4983          |
 
 
 ### Framework versions

all_results.json

@@ -1,14 +1,14 @@
 {
-    "epoch": 2.0,
+    "epoch": 3.0,
     "eval_loss": 0.4982532560825348,
     "eval_runtime": 0.5929,
     "eval_samples": 1,
     "eval_samples_per_second": 1.687,
     "eval_steps_per_second": 1.687,
-    "total_flos": 732097236697088.0,
-    "train_loss": 0.4726724639211012,
-    "train_runtime": 4376.5193,
-    "train_samples": 2913,
-    "train_samples_per_second": 1.331,
-    "train_steps_per_second": 0.042
+    "total_flos": 3793981707124736.0,
+    "train_loss": 0.4249174514315962,
+    "train_runtime": 22985.8992,
+    "train_samples": 10466,
+    "train_samples_per_second": 1.366,
+    "train_steps_per_second": 0.043
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.0,
-    "total_flos": 732097236697088.0,
-    "train_loss": 0.4726724639211012,
-    "train_runtime": 4376.5193,
-    "train_samples": 2913,
-    "train_samples_per_second": 1.331,
-    "train_steps_per_second": 0.042
+    "epoch": 3.0,
+    "total_flos": 3793981707124736.0,
+    "train_loss": 0.4249174514315962,
+    "train_runtime": 22985.8992,
+    "train_samples": 10466,
+    "train_samples_per_second": 1.366,
+    "train_steps_per_second": 0.043
 }