updates GRPOTrainer compatible with trl 0.17 #3969
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
hjh0119
changed the title
Jintao-Huang
approved these changes
Apr 30, 2025
tastelikefeet
approved these changes
Apr 30, 2025
Closed
tastelikefeet
added a commit
to tastelikefeet/swift
that referenced
this pull request
May 6, 2025
* main: fix enable_cache (modelscope#4091) Support ulysses for llm/mllm,dpo/sft (modelscope#4085) update docs (modelscope#4078) feat: support megatron wandb (modelscope#4074) feat: add run name support (modelscope#4072) fix padding_side left (modelscope#4069) bump version support MiMo-7B (modelscope#4067) fix packing eval streaming (modelscope#4066) Support empty think loss scale (modelscope#4065) support qwen3-moe awq (modelscope#4059) Fix grpo eval when gas > 1 (modelscope#4057) fix rollout(modelscope#4055) updates GRPOTrainer compatible with trl 0.17 (modelscope#3969) support Qwen2.5-Omni-3B (modelscope#4052) update wechat (modelscope#4047) # Conflicts: # swift/llm/train/tuner.py
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR type
PR information
update trl 0.16-> 0.17
optional dataset shuffle
huggingface/trl#3334
generate once per effective batch
huggingface/trl#3283
Fix train and eval mode checking
huggingface/trl#3337
optional uvicorn log level for vLLM serve
huggingface/trl#3338
metrics for low and high clipped token probabilities
huggingface/trl#3289
more loss type
huggingface/trl#3256
remove mini_batch
thanks to generate once per effective batch, the mini_batch can be deprecated
Experiment results
Paste your experiment result here(if needed).