Model save

README.md

@@ -1,11 +1,9 @@
 ---
 base_model: allenai/OLMoE-1B-7B-0125-Instruct
-datasets: chenggong1995/NuminaMath-TIR-8000
 library_name: transformers
 model_name: OLMoE-1B-7B-0125-Instruct-grpo-E8-D8000
 tags:
 - generated_from_trainer
-- open-r1
 - trl
 - grpo
 licence: license
@@ -13,7 +11,7 @@ licence: license
 
 # Model Card for OLMoE-1B-7B-0125-Instruct-grpo-E8-D8000
 
-This model is a fine-tuned version of [allenai/OLMoE-1B-7B-0125-Instruct](https://huggingface.co/allenai/OLMoE-1B-7B-0125-Instruct) on the [chenggong1995/NuminaMath-TIR-8000](https://huggingface.co/datasets/chenggong1995/NuminaMath-TIR-8000) dataset.
+This model is a fine-tuned version of [allenai/OLMoE-1B-7B-0125-Instruct](https://huggingface.co/allenai/OLMoE-1B-7B-0125-Instruct).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start
@@ -29,7 +27,7 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/gongc1995-city-university-of-hong-kong/huggingface/runs/0scgpt1n) 
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/gongc1995-city-university-of-hong-kong/huggingface/runs/lnc5ximn) 
 
 
 This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).

all_results.json

@@ -5,9 +5,9 @@
     "eval_samples_per_second": 0.462,
     "eval_steps_per_second": 0.014,
     "total_flos": 0.0,
-    "train_loss": 0.04277147677156233,
-    "train_runtime": 16581.2814,
+    "train_loss": 0.0,
+    "train_runtime": 1.7643,
     "train_samples": 8000,
-    "train_samples_per_second": 3.86,
-    "train_steps_per_second": 0.03
+    "train_samples_per_second": 27205.631,
+    "train_steps_per_second": 210.844
 }

config.json

@@ -28,6 +28,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.49.0",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 50304
 }

tokenizer.json

@@ -1,14 +1,7 @@
 {
   "version": "1.0",
   "truncation": null,
-  "padding": {
-    "strategy": "BatchLongest",
-    "direction": "Left",
-    "pad_to_multiple_of": null,
-    "pad_id": 50280,
-    "pad_type_id": 0,
-    "pad_token": "<pad>"
-  },
+  "padding": null,
   "added_tokens": [
     {
       "id": 0,

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 0.04277147677156233,
-    "train_runtime": 16581.2814,
+    "train_loss": 0.0,
+    "train_runtime": 1.7643,
     "train_samples": 8000,
-    "train_samples_per_second": 3.86,
-    "train_steps_per_second": 0.03
+    "train_samples_per_second": 27205.631,
+    "train_steps_per_second": 210.844
 }

trainer_state.json

@@ -1526,27 +1526,19 @@
       "step": 495
     },
     {
-      "clip_ratio": 0.0,
-      "completion_length": 788.9921875,
       "epoch": 7.992,
-      "kl": 1.58984375,
-      "reward": 1.0631510764360428,
-      "reward_std": 0.9985880479216576,
-      "rewards/accuracy_reward": 0.09635416883975267,
-      "rewards/format_reward": 0.4479166753590107,
-      "rewards/tag_count_reward": 0.5188802257180214,
       "step": 496,
       "total_flos": 0.0,
-      "train_loss": 0.04277147677156233,
-      "train_runtime": 16581.2814,
-      "train_samples_per_second": 3.86,
-      "train_steps_per_second": 0.03
+      "train_loss": 0.0,
+      "train_runtime": 1.7643,
+      "train_samples_per_second": 27205.631,
+      "train_steps_per_second": 210.844
     }
   ],
   "logging_steps": 5,
-  "max_steps": 496,
+  "max_steps": 372,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 8,
+  "num_train_epochs": 6,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7d2f4c875dace7ca6ceda95d7c63751bfe39ce057d75980444e10a701401db79
+oid sha256:c41642c5de67e0949b20ca2984a6e09f10bc7eaf1389f41ec835f9c6be67aefa
 size 8056