Upload train_arithmetic_v2.py with huggingface_hub

train_arithmetic_v2.py

@@ -16,7 +16,7 @@ import re
 import random
 import torch
 from datasets import Dataset
-from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
+from transformers import AutoModelForCausalLM, AutoTokenizer
 from trl import GRPOConfig, GRPOTrainer
 
 # ============================================================================
@@ -243,14 +243,6 @@ def main():
         save_steps=MAX_STEPS,
         push_to_hub=False,
         report_to="none",
-        # Force EOS in generation
-        generation_config=GenerationConfig(
-            max_new_tokens=30,
-            do_sample=True,
-            temperature=0.7,
-            pad_token_id=tokenizer.eos_token_id,
-            eos_token_id=tokenizer.eos_token_id,
-        ),
     )
     
     # Eval callback