[megatron, ckpt] feat: add Qwen3.5 Megatron-FSDP SFT example#6352
Draft
conver334 wants to merge 6 commits into
Draft
[megatron, ckpt] feat: add Qwen3.5 Megatron-FSDP SFT example#6352conver334 wants to merge 6 commits into
conver334 wants to merge 6 commits into
Conversation
Contributor
There was a problem hiding this comment.
Code Review
This pull request introduces a new shell script for running Qwen3.5-35B-A3B SFT using Megatron-FSDP on 32 GPUs, providing a comprehensive configuration for distributed training. Additionally, it updates the Megatron checkpoint manager to use default=str in json.dump when saving transformer configurations, ensuring that non-serializable objects do not cause errors during checkpointing. I have no feedback to provide.
Signed-off-by: conver334 <conver334@gmail.com>
a35091e to
056baa3
Compare
Signed-off-by: conver334 <conver334@gmail.com>
Set data.use_dynamic_bsz=True together with engine.use_remove_padding=True and model.use_remove_padding=True so the 32-GPU Qwen3.5 Megatron-FSDP SFT example runs in THD mode by default. Signed-off-by: conver334 <conver334@gmail.com>
6deee1e to
838b782
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
Adds a GSM8K SFT example for Qwen3.5-35B-A3B with Megatron-FSDP on 32 GPUs.
The example documents the required latest Megatron-Bridge, Megatron-LM dev branch, and runtime Python package dependencies.
The required Megatron-FSDP support has already been merged into
Megatron-LMdev branch (NVIDIA/Megatron-LM#4799) andMegatron-Bridgemain (NVIDIA-NeMo/Megatron-Bridge#3746)Checklist Before Starting
[{modules}] {type}: {description}Design & Code Changes
Checklist Before Submitting
and Megatron-LM changes.