From f69cc7c178d5bd47cd9bc5cf4c2f41530b878aae Mon Sep 17 00:00:00 2001 From: Jinzhen Lin Date: Wed, 23 Sep 2026 23:22:45 +0800 Subject: [PATCH] [Quantization][Bugfix] Bump humming-kernels to 0.1.16 (#58054) Signed-off-by: jinzhen.ljz Co-authored-by: Codex Co-authored-by: Misha Goin --- requirements/cuda.txt | 2 +- tests/evals/gsm8k/configs/humming/Qwen3-0.6B-MXFP8-humming.yaml | 1 - .../gsm8k/configs/humming/Qwen3-30B-A3B-FP8-block-humming.yaml | 1 - 3 files changed, 1 insertion(+), 3 deletions(-) diff --git a/requirements/cuda.txt b/requirements/cuda.txt index 72609d8983..59e5c4dcc8 100644 --- a/requirements/cuda.txt +++ b/requirements/cuda.txt @@ -33,4 +33,4 @@ quack-kernels==0.6.5 # Required by MSA tokenspeed-mla==0.1.8; platform_system == "Linux" # Humming kernels for quantization gemm -humming-kernels[cu13]==0.1.15 +humming-kernels[cu13]==0.1.16 diff --git a/tests/evals/gsm8k/configs/humming/Qwen3-0.6B-MXFP8-humming.yaml b/tests/evals/gsm8k/configs/humming/Qwen3-0.6B-MXFP8-humming.yaml index b83d9e6a9e..c0bb8d3240 100644 --- a/tests/evals/gsm8k/configs/humming/Qwen3-0.6B-MXFP8-humming.yaml +++ b/tests/evals/gsm8k/configs/humming/Qwen3-0.6B-MXFP8-humming.yaml @@ -3,6 +3,5 @@ accuracy_threshold: 0.39 num_questions: 1319 num_fewshot: 5 server_args: >- - --enforce-eager --max-model-len 8192 --linear-backend humming diff --git a/tests/evals/gsm8k/configs/humming/Qwen3-30B-A3B-FP8-block-humming.yaml b/tests/evals/gsm8k/configs/humming/Qwen3-30B-A3B-FP8-block-humming.yaml index 8ca4777ce5..357e7a0257 100644 --- a/tests/evals/gsm8k/configs/humming/Qwen3-30B-A3B-FP8-block-humming.yaml +++ b/tests/evals/gsm8k/configs/humming/Qwen3-30B-A3B-FP8-block-humming.yaml @@ -3,7 +3,6 @@ accuracy_threshold: 0.86 num_questions: 1319 num_fewshot: 5 server_args: >- - --enforce-eager --max-model-len 8192 --moe-backend humming --linear-backend humming