diff --git a/train_grpo.py b/train_grpo.py index 44b750b..7c7ada7 100644 --- a/train_grpo.py +++ b/train_grpo.py @@ -88,7 +88,7 @@ def agentic_generate( ): # Create agent with appropriate adapter based on tokenizer tokenizer_name = tokenizer.name_or_path.lower() - if "deepseek-r1-distill" in tokenizer_name: + if "deepseek-ai/deepseek-r1-distill" in tokenizer_name: adapter = R1DistilTokenizerAdapter() elif "llama" in tokenizer_name: adapter = LlamaTokenizerAdapter()