Conversation
The NemotronH backbone already routes lora_params to attention, both Mamba projections, the shared expert and the MoE latent projections. The NemotronH_Nano_VL_V2 wrapper forwards lora_params but never exposed lora_config()/lora_request(), and callers resolve those on the outer class, so every documented LoRA entry point hit AttributeError. Forward both to NemotronHForCausalLM, as Phi-4-MM does. Add the first multi-LoRA coverage for a multimodal model. Adapters are fabricated, since none is published for this family. CPU tests assert every adapter key reaches a module: an unrecognized key is warned about and skipped, so a mis-built adapter loads clean and stays inert. The GPU test runs three adapters of differing rank plus a no-LoRA row in one batch, every row carrying an image, and asserts the no-LoRA row is untouched and the adapters diverge from base and from each other. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Yueh-Ting Chen <yuehtingc@nvidia.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Enable multi-LoRA on Nemotron 3.5 Super VL.
Working combination: 2×B200, TP2,
nemotron_3.5_super_nvfp4_experts_no_fp8_kv:Test Coverage
The one test that fails without the source change:
test_vl_wrapper_exposes_lora_config(tests/unittest/_torch/modules/tests_lora_modules/test_nemotron35_vl_lora_adapter.py:232) — on base,NemotronH_Nano_VL_V2.lora_config()raisesAttributeError.The CPU tests run in the existing unit-test stage, no test-list change. The GPU test is not in a test-db list yet: it needs the 3.5 Super checkpoint, which is not staged under
LLM_MODELS_ROOT, so it skips there today and runs viaNEMOTRON35_VL_MODEL_DIR.PR Checklist
Please review the following before submitting your PR:
PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
If PR introduces API changes, an appropriate PR label is added - either
api-compatibleorapi-breaking. Forapi-breaking, includeBREAKINGin the PR title.Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.
GitHub Bot Help
To see a list of available CI bot commands, please comment
/bot help.🤖 Generated with Claude Code