🫣 [GRPO] add cache_implementation option in GRPO

trl/trainer/grpo_config.py

@@ -72,6 +72,8 @@ class GRPOConfig(TrainingArguments):
             Float that penalizes new tokens based on whether they appear in the prompt and the generated text so far.
             Values > `1.0` encourage the model to use new tokens, while values < `1.0` encourage the model to repeat
             tokens.
+        cache_implementation (`str` or `None`, *optional*, defaults to `None`):
+            Implementation of the cache method for faster generation when use_vllm is set to False.
 
         > Parameters that control generation acceleration powered by vLLM
 
@@ -217,6 +219,10 @@ class GRPOConfig(TrainingArguments):
             "to repeat tokens."
         },
     )
+    cache_implementation: Optional[str] = field(
+        default=None,
+        metadata={"help": "Implementation of the cache method for faster generation when use_vllm is set to False."},
+    )
 
     # Parameters that control generation acceleration powered by vLLM
     use_vllm: Optional[bool] = field(

trl/trainer/grpo_trainer.py

@@ -548,6 +548,7 @@ def new_group_context():
                 top_k=args.top_k,
                 min_p=args.min_p,
                 repetition_penalty=args.repetition_penalty,
+                cache_implementation=args.cache_implementation,
             )
 
         # Gradient accumulation requires scaled loss. Normally, loss scaling in the parent class depends on whether the