We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 97d646b commit d88badaCopy full SHA for d88bada
swift/trainers/rlhf_trainer/grpo_trainer.py
@@ -172,7 +172,11 @@ def __init__(self,
172
self.max_completion_length = args.max_completion_length
173
self.completion_length_limit_scope = args.completion_length_limit_scope
174
model.warnings_issued['estimate_tokens'] = True
175
- kwargs['data_collator'] = lambda features: features
+
176
+ def data_collator(features): # No data collation is needed in GRPO
177
+ return features
178
179
+ kwargs['data_collator'] = data_collator
180
self.shuffle_dataset = args.dataset_shuffle
181
182
self.use_vllm = args.use_vllm
0 commit comments