From af28c171f5e786a67206124dcbf925c501128e50 Mon Sep 17 00:00:00 2001 From: Anerudhan Gopal Date: Thu, 15 Oct 2026 17:04:24 -0700 Subject: [PATCH] [Dependency] Upgrade FlashInfer to 0.7.2 and restore KDA auto backend (#61913) Signed-off-by: Anerudhan Gopal Co-authored-by: OpenAI Codex --- requirements/cuda.txt | 2 +- vllm/models/kimi_k3/nvidia/kda.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/requirements/cuda.txt b/requirements/cuda.txt index ad2bc63df1..a990dda387 100644 --- a/requirements/cuda.txt +++ b/requirements/cuda.txt @@ -10,7 +10,7 @@ torchaudio==2.11.0 torchvision==0.29.0 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version torchcodec >= 0.14 PyNvVideoCodec==2.0.4 -flashinfer-python==0.7.1 +flashinfer-python==0.7.2 apache-tvm-ffi==0.1.11 tilelang==0.1.12 nvidia-cudnn-frontend>=1.19.1 diff --git a/vllm/models/kimi_k3/nvidia/kda.py b/vllm/models/kimi_k3/nvidia/kda.py index 284e4585c6..c8cfd8cf45 100644 --- a/vllm/models/kimi_k3/nvidia/kda.py +++ b/vllm/models/kimi_k3/nvidia/kda.py @@ -456,7 +456,7 @@ def _flashinfer_kda_prefill( beta_is_logit=True, seq_order=seq_order, prefill_workspace=prefill_workspace, - backend="cute-dsl", + backend="auto", ) return output, initial_state