End of training

README.md

@@ -1,9 +1,11 @@
 ---
 base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
+datasets: DigitalLearningGmbH/MATH-lighteval
 library_name: transformers
 model_name: DeepSeek-R1-Distill-Qwen-7B-GRPO_Math
 tags:
 - generated_from_trainer
+- open-r1
 - trl
 - grpo
 licence: license
@@ -11,7 +13,7 @@ licence: license
 
 # Model Card for DeepSeek-R1-Distill-Qwen-7B-GRPO_Math
 
-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).
+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 [DigitalLearningGmbH/MATH-lighteval](https://huggingface.co/datasets/DigitalLearningGmbH/MATH-lighteval) dataset.
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start

all_results.json

@@ -1,9 +1,9 @@
 {
-    "eval_loss": -7.450580596923828e-09,
-    "eval_runtime": 18.8159,
-    "eval_samples": 7,
-    "eval_samples_per_second": 0.372,
-    "eval_steps_per_second": 0.053,
+    "eval_loss": 0.0013432882260531187,
+    "eval_runtime": 7436.9567,
+    "eval_samples": 5000,
+    "eval_samples_per_second": 0.672,
+    "eval_steps_per_second": 0.096,
     "total_flos": 0.0,
     "train_loss": 0.001102253468012187,
     "train_runtime": 12358.4402,

config.json

@@ -23,7 +23,7 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.49.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "use_mrope": false,
   "use_sliding_window": false,
   "vocab_size": 152064

eval_results.json

@@ -1,7 +1,7 @@
 {
-    "eval_loss": -7.450580596923828e-09,
-    "eval_runtime": 18.8159,
-    "eval_samples": 7,
-    "eval_samples_per_second": 0.372,
-    "eval_steps_per_second": 0.053
+    "eval_loss": 0.0013432882260531187,
+    "eval_runtime": 7436.9567,
+    "eval_samples": 5000,
+    "eval_samples_per_second": 0.672,
+    "eval_steps_per_second": 0.096
 }