FreeToken brings datacenter-scale model serving to your desktop. Run massive models locally, fast and efficiently.
fix(install): pin flashinfer packages to one version (#545)
Xiaoze Fan71eb1f01660f
diff --git a/install.sh b/install.shindex 60b4439c..2a9a3dec 100755--- a/install.sh+++ b/install.sh@@ -24,6 +24,8 @@# FREETOKEN_PY_VERSION python for the venv (default: 3.12 — must match the wheel tag)# FREETOKEN_BIN_DIR where to symlink `ft` (default: ~/.local/bin)# FREETOKEN_ENV_DIR environment.d dir (default: ~/.config/environment.d)+# FREETOKEN_FLASHINFER_VERSION one version for flashinfer-python/-cubin/-jit-cache+# (default: $DEFAULT_FLASHINFER_VERSION)## NOTE: common TVM FFI kernels come from the kernel-cache wheel. A working CUDA# toolkit (nvcc) is still needed when falling back to JIT for an uncovered kernel@@ -32,6 +34,7 @@ set -euo pipefailDEFAULT_WHEEL_URL="" # filled in once GitHub Releases are liveDEFAULT_KERNEL_CACHE_WHEEL_URL="" # filled in once GitHub Releases are live+DEFAULT_FLASHINFER_VERSION="0.6.18.post1"FT_HOME="${FREETOKEN_HOME:-$HOME/.freetoken}"VENV="$FT_HOME/venv"@@ -40,6 +43,7 @@ BIN_DIR="${FREETOKEN_BIN_DIR:-$HOME/.local/bin}"ENV_DIR="${FREETOKEN_ENV_DIR:-$HOME/.config/environment.d}"WHEEL="${FREETOKEN_WHEEL:-$DEFAULT_WHEEL_URL}"KERNEL_CACHE_WHEEL="${FREETOKEN_KERNEL_CACHE_WHEEL:-$DEFAULT_KERNEL_CACHE_WHEEL_URL}"+FLASHINFER_VERSION="${FREETOKEN_FLASHINFER_VERSION:-$DEFAULT_FLASHINFER_VERSION}"# --yes / -y (or FREETOKEN_ASSUME_YES=1): run non-interactively — in particular, bootstrap uv# without asking. FreeToken Desktop's in-app installer passes --yes (its stdout is piped into a@@ -211,10 +215,12 @@ mkdir -p "$FT_HOME"# breaking driver-only on a box with no CUDA toolkit. flashinfer-cubin + flashinfer-jit-cache ship# those kernels PREBUILT (multi-arch), so nothing compiles at runtime. They live on flashinfer's# own index (cubin arch-agnostic; jit-cache per-cuNNN). Large (~2 GiB) but downloaded once.+# flashinfer refuses to import unless all three share one version, so pin them together.INSTALL_WHEELS=("${WHEEL}[accel]"- flashinfer-cubin- flashinfer-jit-cache+ "flashinfer-python==$FLASHINFER_VERSION"+ "flashinfer-cubin==$FLASHINFER_VERSION"+ "flashinfer-jit-cache==$FLASHINFER_VERSION""$KERNEL_CACHE_WHEEL")CU_INDEX_ARGS=(diff --git a/pyproject.toml b/pyproject.tomlindex 6b247a2c..25cb204b 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -76,7 +76,7 @@ dev = ["pytest>=6.0"]# Native fused-kernel packages. Without them the runtime falls back to the pure-Triton# kernels in freetoken.kernel.triton (and the 'triton' attention backend). Install# `freetoken[accel]` for the full native fast path.-fi = ["flashinfer-python[cu13]>=0.6,<0.7"]+fi = ["flashinfer-python[cu13]==0.6.18.post1"]# renamed from sgl-kernel at 0.4; still imports as `sgl_kernel`, so never co-install bothsgl = ["sglang-kernel==0.4.5"]accel = ["freetoken[fi,sgl]"]