feat: Add support for gemma-3 #298
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
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
terrykong
reviewed
Apr 30, 2025
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
terrykong
reviewed
May 6, 2025
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Signed-off-by: Yi-Fu Wu 8000 <yifu.wu@gmail.com>
Open
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
terrykong
approved these changes
May 8, 2025
examples/configs/recipes/llm/grpo-gemma3-1b-it-1n8g-fsdp2tp1.yaml
Outdated
Show resolved
Hide resolved
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Outdated suggestions cannot be applied.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.
What does this PR do ?
Add support for gemma-3 models
Issues
Closes #236
Usage
Since Gemma-3 models have tied weights for all models, the TP and SP strategies are configured to not parallelize the embed_tokens and lm_head layers. To run, use the
NRL_SKIP_TIED_WEIGHT_CHECK=1flag. Additionally, use++policy.generation.vllm_cfg.load_format=autobecause vllm has a bug using dummy weights with gemma-3 models.Before your PR is "Ready for review"
Pre checks:
Additional Information
Sample runs for 1b and 4b: https://wandb.ai/nvidia/grpo-dev-yifu/workspace?nw=7ii089mxlc5
Sample runs using TP/SP: https://wandb.ai/nvidia/grpo-dev-yifu?nw=serntftinh