chore: bump flashinfer version to 0.6.6 (#20480)
Co-authored-by: sglang-bot <sglang-bot@users.noreply.github.com>
This commit is contained in:
@@ -19,7 +19,7 @@ ARG PIP_DEFAULT_INDEX
|
||||
ARG UBUNTU_MIRROR
|
||||
ARG GITHUB_ARTIFACTORY=github.com
|
||||
ARG INSTALL_FLASHINFER_JIT_CACHE=0
|
||||
ARG FLASHINFER_VERSION=0.6.4
|
||||
ARG FLASHINFER_VERSION=0.6.6
|
||||
ARG MOONCAKE_VERSION=0.3.9
|
||||
#if need other arg please add in MOONCAKE_COMPILE_ARG
|
||||
ARG MOONCAKE_COMPILE_ARG="-DUSE_HTTP=ON -DUSE_MNNVL=ON -DUSE_CUDA=ON -DWITH_EP=ON"
|
||||
|
||||
@@ -27,8 +27,8 @@ dependencies = [
|
||||
"datasets",
|
||||
"einops",
|
||||
"fastapi",
|
||||
"flashinfer_python==0.6.4", # keep it aligned with jit-cache version in Dockerfile
|
||||
"flashinfer_cubin==0.6.4",
|
||||
"flashinfer_python==0.6.6", # keep it aligned with jit-cache version in Dockerfile
|
||||
"flashinfer_cubin==0.6.6",
|
||||
"gguf",
|
||||
"hf_transfer",
|
||||
"huggingface_hub",
|
||||
|
||||
@@ -1135,7 +1135,7 @@ def _set_envs_and_config(server_args: ServerArgs):
|
||||
if server_args.attention_backend == "flashinfer":
|
||||
assert_pkg_version(
|
||||
"flashinfer_python",
|
||||
"0.6.4",
|
||||
"0.6.6",
|
||||
"Please uninstall the old version and "
|
||||
"reinstall the latest version by following the instructions "
|
||||
"at https://docs.flashinfer.ai/installation.html.",
|
||||
|
||||
@@ -1132,7 +1132,7 @@ def check_pkg_version_at_least(pkg: str, min_version: str) -> bool:
|
||||
|
||||
Args:
|
||||
pkg: Package name (distribution name, e.g., "flashinfer-python")
|
||||
min_version: Minimum version required (e.g., "0.6.4")
|
||||
min_version: Minimum version required (e.g., "0.6.6")
|
||||
|
||||
Returns:
|
||||
True if package is installed and version >= min_version, False otherwise
|
||||
|
||||
@@ -5,7 +5,7 @@ set -euxo pipefail
|
||||
# Set up environment variables
|
||||
IS_BLACKWELL=${IS_BLACKWELL:-0}
|
||||
CU_VERSION="cu129"
|
||||
FLASHINFER_VERSION=0.6.4
|
||||
FLASHINFER_VERSION=0.6.6
|
||||
OPTIONAL_DEPS="${1:-}"
|
||||
|
||||
# Detect system architecture
|
||||
|
||||
Reference in New Issue
Block a user