feat: add gradient checkpointing in megatron backend #944
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
eric-haibin-lin
approved these changes
Apr 6, 2025
yuchenwang3
pushed a commit
to yuchenwang3/verl
that referenced
this pull request
Apr 25, 2025
…engine#944) ### Changes Add gradient checkpointing (aka `activation recomputation`) config and support from Megatron core (https://github.com/NVIDIA/Megatron-LM/blob/b7ec711cf66cf500b98d8783f2c7f3c3a7d5ba31/megatron/core/transformer/transformer_config.py#L208-L233) to make activation checkpointing more efficient for LLMs with 20B+ parameters. ``` gradient_checkpointing_kwargs: activations_checkpoint_method: null activations_checkpoint_granularity: null activations_checkpoint_num_layers: null ``` ### Test Tested on loading Qwen7b/32b of 16k input prompts and bypass the OOM issues after adding gradient checkpointing. ### Next Step Add one `ppo_trainer for megatron` doc to explain the config details in https://verl.readthedocs.io/en/latest/examples/config.html
histmeisah
pushed a commit
to SJTU-IAAR/verl
that referenced
this pull request
Apr 27, 2025
…engine#944) ### Changes Add gradient checkpointing (aka `activation recomputation`) config and support from Megatron core (https://github.com/NVIDIA/Megatron-LM/blob/b7ec711cf66cf500b98d8783f2c7f3c3a7d5ba31/megatron/core/transformer/transformer_config.py#L208-L233) to make activation checkpointing more efficient for LLMs with 20B+ parameters. ``` gradient_checkpointing_kwargs: activations_checkpoint_method: null activations_checkpoint_granularity: null activations_checkpoint_num_layers: null ``` ### Test Tested on loading Qwen7b/32b of 16k input prompts and bypass the OOM issues after adding gradient checkpointing. ### Next Step Add one `ppo_trainer for megatron` doc to explain the config details in https://verl.readthedocs.io/en/latest/examples/config.html
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.
Changes
Add gradient checkpointing (aka
activation recomputation) config and support from Megatron core (https://github.com/NVIDIA/Megatron-LM/blob/b7ec711cf66cf500b98d8783f2c7f3c3a7d5ba31/megatron/core/transformer/transformer_config.py#L208-L233) to make activation checkpointing more efficient for LLMs with 20B+ parameters.Test
Tested on loading Qwen7b/32b of 16k input prompts and bypass the OOM issues after adding gradient checkpointing.
Next Step
Add one
ppo_trainer for megatrondoc to explain the config details in https://verl.readthedocs.io/en/latest/examples/config.html