Repository navigation
Conversation
nvfp4_weight_only now also quantizes routed experts: the online NVFP4 MoE method accepts activation=null and runs the experts on the W4A16 NVFP4 MoE backends (Marlin, Humming). One layer kind alone stays available via --quantization-config.linear or --quantization-config.moe nvfp4_weight_only. Signed-off-by: Stefano Castagnetta <scastagnetta@nvidia.com>
stecasta
requested review from
AndreasKaratzas,
ProExpertProg,
WoosukKwon,
houseroad,
mgoin,
pavanimajety,
robertgshaw2-redhat,
tlrmchlsmth,
yewentao256,
youkaichao and
zyongye
as code owners
November 16, 2026 20:54
Contributor
|
Documentation preview: https://vllm--68018.org.readthedocs.build/en/68018/ |
Contributor
|
This pull request has merge conflicts that must be resolved before it can be |
Signed-off-by: Stefano Castagnetta <scastagnetta@nvidia.com> # Conflicts: # vllm/model_executor/layers/quantization/online/nvfp4.py
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Purpose
Follow-up to #67006:
nvfp4_weight_onlynow also quantizes routed experts, as suggested in review there. The online NVFP4 MoE method acceptsactivation: nulland runs the experts on the existing W4A16 NVFP4 MoE backends (Marlin, Humming). No new kernels.--quantization-config.linear nvfp4_weight_onlyor--quantization-config.moe nvfp4_weight_onlyquantize only one layer kind.nvfp4_per_tokenis unchanged.Note: #67136 also edits
_quantize_moe_weight_to_nvfp4; whichever lands second rebases.Test Plan
pytest tests/quantization/test_online.py tests/quantization/test_quantization_config_args.py, stock vs this PR, on GB10, GB300 and RTX PRO 6000.tests/evals/gsm8k/gsm8k_eval.py(1319 questions).Test Result
Tests: no regressions on any of the three GPUs; the new tests pass. Marlin is auto-selected for the W4A16 experts on all three.
Decode tok/s at c=1,
--quantization <arm>:fp8_per_channelnvfp4_weight_only, main (experts BF16)nvfp4_weight_only, this PR+38% (Spark) and +16% (RTX PRO 6000) decode over the current shorthand, 3.2x less weight memory, accuracy unchanged.
This PR was written with AI assistance; I reviewed every line and ran the tests and benchmarks above. Searched open PRs for online weight-only NVFP4 MoE: none besides the overlap noted above.