Skip to content

Use pre-built wheel for flash-attn on Linux & Resolve version mismatch for torchaudio #1848

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
2vibeornot2vibe wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Use pre-built wheel for flash-attn on Linux & Resolve version mismatch for torchaudio #1848

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3875dd0
Use pre-built wheel for flash-attn on Linux
2vibeornot2vibe May 30, 2026
4bd7e33
Specify torchvision & torchaudio versions and update CUDA 12.8 config…
2vibeornot2vibe May 30, 2026
57499eb
Use Python 3.11 instead of 3.10 as default for pre-Turin cards (teste…
2vibeornot2vibe May 30, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
122 changes: 61 additions & 61 deletions setup_config.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,26 +4,26 @@
"3.10": { "ver": "3.10.9", "label": "Python 3.10.9 (Legacy)" },
"3.11": { "ver": "3.11.14", "label": "Python 3.11.14 (Recommended)" }
},
"torch": {
"cu128": {
"label": "Torch 2.7.1 + CUDA 12.8",
"cmd": "torch==2.7.1 torchvision==0.22.1 torchaudio==2.7.1 --index-url https://download.pytorch.org/whl/cu128"
},
"cu130": {
"label": "Torch 2.10.0 + CUDA 13.0",
"cmd": "torch==2.10.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu130"
},
"rocm65": {
"label": "ROCm 6.5 (TheRock)",
"cmd": {
"win": "https://github.com/scottt/rocm-TheRock/releases/download/v6.5.0rc-pytorch-gfx110x/torch-2.7.0a0+rocm_git3f903c3-cp311-cp311-win_amd64.whl https://github.com/scottt/rocm-TheRock/releases/download/v6.5.0rc-pytorch-gfx110x/torchvision-0.22.0+9eb57cd-cp311-cp311-win_amd64.whl https://github.com/scottt/rocm-TheRock/releases/download/v6.5.0rc-pytorch-gfx110x/torchaudio-2.7.0a0+52638ef-cp311-cp311-win_amd64.whl",
"linux": "torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.2"
}
},
"mps": {
"label": "Torch (MPS / Apple Silicon)",
"cmd": "torch torchvision torchaudio"
}
"torch": {
"cu128": {
"label": "Torch 2.10.0 + CUDA 12.8",
"cmd": "torch==2.10.0+cu128 torchvision==0.25.0+cu128 torchaudio==2.10.0+cu128 --index-url https://download.pytorch.org/whl/cu128"
},
"cu130": {
"label": "Torch 2.10.0 + CUDA 13.0",
"cmd": "torch==2.10.0 torchvision==0.25.0 torchaudio==2.10.0 --index-url https://download.pytorch.org/whl/cu130"
},
"rocm65": {
"label": "ROCm 6.5 (TheRock)",
"cmd": {
"win": "https://github.com/scottt/rocm-TheRock/releases/download/v6.5.0rc-pytorch-gfx110x/torch-2.7.0a0+rocm_git3f903c3-cp311-cp311-win_amd64.whl https://github.com/scottt/rocm-TheRock/releases/download/v6.5.0rc-pytorch-gfx110x/torchvision-0.22.0+9eb57cd-cp311-cp311-win_amd64.whl https://github.com/scottt/rocm-TheRock/releases/download/v6.5.0rc-pytorch-gfx110x/torchaudio-2.7.0a0+52638ef-cp311-cp311-win_amd64.whl",
"linux": "torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.2"
}
},
"mps": {
"label": "Torch (MPS / Apple Silicon)",
"cmd": "torch torchvision torchaudio"
}
},
"triton": {
"v33": {
Expand All @@ -48,11 +48,11 @@
}
}
},
"sage": {
"v1": { "label": "Sage Attention 1.0.6", "cmd": "sageattention==1.0.6" },
"v211": {
"label": "Sage Attention 2.1.1",
"cmd": {
"sage": {
"v1": { "label": "Sage Attention 1.0.6", "cmd": "sageattention==1.0.6" },
"v211": {
"label": "Sage Attention 2.1.1",
"cmd": {
"win": "https://github.com/woct0rdho/SageAttention/releases/download/v2.1.1-windows/sageattention-2.1.1+cu126torch2.6.0-cp310-cp310-win_amd64.whl",
"linux": "pip install \"setuptools<=75.8.2\" ninja wheel --force-reinstall && git clone https://github.com/thu-ml/SageAttention && pip install --no-build-isolation -e SageAttention"
}
Expand All @@ -69,31 +69,31 @@
"cmd": {
"win": "https://github.com/woct0rdho/SageAttention/releases/download/v2.2.0-windows.post4/sageattention-2.2.0+cu130torch2.9.0andhigher.post4-cp39-abi3-win_amd64.whl",
"linux": "pip install \"setuptools<=75.8.2\" ninja wheel --force-reinstall && git clone https://github.com/thu-ml/SageAttention && pip install --no-build-isolation -e SageAttention"
}
}
},
"sparge": {
"v010_cu128": {
"label": "Sparge Attention 0.1.0 (CUDA 12.8 / Torch 2.7.1)",
"cmd": {
"win": "https://github.com/woct0rdho/SpargeAttn/releases/download/v0.1.0-windows.post3/spas_sage_attn-0.1.0%2Bcu128torch2.7.1.post3-cp39-abi3-win_amd64.whl",
"linux": "{pip} ninja wheel packaging && {pip} --no-build-isolation git+https://github.com/woct0rdho/SpargeAttn.git"
}
},
"v010_cu13": {
"label": "Sparge Attention 0.1.0 (CUDA 13 / Torch 2.9+ ABI)",
"cmd": {
"win": "https://github.com/woct0rdho/SpargeAttn/releases/download/v0.1.0-windows.post4/spas_sage_attn-0.1.0%2Bcu130torch2.9.0andhigher.post4-cp39-abi3-win_amd64.whl",
"linux": "{pip} ninja wheel packaging && {pip} --no-build-isolation git+https://github.com/woct0rdho/SpargeAttn.git"
}
}
},
"flash": {
"v210": {
"label": "Flash Attention 2.8.x",
"cmd": {
}
}
},
"sparge": {
"v010_cu128": {
"label": "Sparge Attention 0.1.0 (CUDA 12.8 / Torch 2.7.1)",
"cmd": {
"win": "https://github.com/woct0rdho/SpargeAttn/releases/download/v0.1.0-windows.post3/spas_sage_attn-0.1.0%2Bcu128torch2.7.1.post3-cp39-abi3-win_amd64.whl",
"linux": "{pip} ninja wheel packaging && {pip} --no-build-isolation git+https://github.com/woct0rdho/SpargeAttn.git"
}
},
"v010_cu13": {
"label": "Sparge Attention 0.1.0 (CUDA 13 / Torch 2.9+ ABI)",
"cmd": {
"win": "https://github.com/woct0rdho/SpargeAttn/releases/download/v0.1.0-windows.post4/spas_sage_attn-0.1.0%2Bcu130torch2.9.0andhigher.post4-cp39-abi3-win_amd64.whl",
"linux": "{pip} ninja wheel packaging && {pip} --no-build-isolation git+https://github.com/woct0rdho/SpargeAttn.git"
}
}
},
"flash": {
"v210": {
"label": "Flash Attention 2.8.x",
"cmd": {
"win": "https://github.com/deepbeepmeep/kernels/releases/download/Flash2/flash_attn-2.8.3-cp311-cp311-win_amd64.whl",
"linux": "flash-attn"
"linux": "https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.9.0/flash_attn-2.8.3+cu130torch2.10-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl"
}
}
},
Expand Down Expand Up @@ -128,15 +128,15 @@
}
}
},
"gpu_profiles": {
"GTX_10": { "python": "3.10", "torch": "cu128", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [] },
"RTX_20": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v1", "sparge": null, "flash": "v210", "kernels": ["nunchaku_cu13", "gguf"] },
"RTX_30": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v220_cu13", "sparge": "v010_cu13", "flash": "v210", "kernels": ["nunchaku_cu13", "gguf"] },
"RTX_40": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v220_cu13", "sparge": "v010_cu13", "flash": "v210", "kernels": ["nunchaku_cu13", "gguf"] },
"RTX_50": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v220_cu13", "sparge": "v010_cu13", "flash": "v210", "kernels": ["nunchaku_cu13","light2xv", "gguf"] },
"AMD_GFX110X": { "python": "3.11", "torch": "rocm65", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "env": {"HSA_OVERRIDE_GFX_VERSION": "11.0.0"} },
"AMD_GFX1151": { "python": "3.11", "torch": "rocm65", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "env": {"HSA_OVERRIDE_GFX_VERSION": "11.5.1"} },
"AMD_GFX1201": { "python": "3.11", "torch": "rocm65", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "env": {"HSA_OVERRIDE_GFX_VERSION": "12.0.1"} },
"MPS": { "python": "3.11", "torch": "mps", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "attention": "sdpa" }
}
}
"gpu_profiles": {
"GTX_10": { "python": "3.11", "torch": "cu128", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [] },
"RTX_20": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v1", "sparge": null, "flash": "v210", "kernels": ["nunchaku_cu13", "gguf"] },
"RTX_30": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v220_cu13", "sparge": "v010_cu13", "flash": "v210", "kernels": ["nunchaku_cu13", "gguf"] },
"RTX_40": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v220_cu13", "sparge": "v010_cu13", "flash": "v210", "kernels": ["nunchaku_cu13", "gguf"] },
"RTX_50": { "python": "3.11", "torch": "cu130", "triton": "latest", "sage": "v220_cu13", "sparge": "v010_cu13", "flash": "v210", "kernels": ["nunchaku_cu13","light2xv", "gguf"] },
"AMD_GFX110X": { "python": "3.11", "torch": "rocm65", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "env": {"HSA_OVERRIDE_GFX_VERSION": "11.0.0"} },
"AMD_GFX1151": { "python": "3.11", "torch": "rocm65", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "env": {"HSA_OVERRIDE_GFX_VERSION": "11.5.1"} },
"AMD_GFX1201": { "python": "3.11", "torch": "rocm65", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "env": {"HSA_OVERRIDE_GFX_VERSION": "12.0.1"} },
"MPS": { "python": "3.11", "torch": "mps", "triton": null, "sage": null, "sparge": null, "flash": null, "kernels": [], "attention": "sdpa" }
}
}