Promote latest kernel artifacts to main
Browse files- README.md +0 -9
- build/torch211-cxx11-cu130-aarch64-linux/__init__.py +53 -0
- build/torch211-cxx11-cu130-aarch64-linux/{_fp4_gemm_cuda_7781728.abi3.so → _fp4_gemm_cuda_b46a817.abi3.so} +2 -2
- build/torch211-cxx11-cu130-aarch64-linux/_ops.py +3 -3
- build/torch211-cxx11-cu130-aarch64-linux/metadata.json +5 -5
- build/torch212-cxx11-cu130-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so} +1 -1
- build/torch212-cxx11-cu130-x86_64-linux/_ops.py +3 -3
- build/torch212-cxx11-cu130-x86_64-linux/metadata.json +4 -4
- build/torch212-cxx11-cu132-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so} +1 -1
- build/torch212-cxx11-cu132-x86_64-linux/_ops.py +3 -3
- build/torch212-cxx11-cu132-x86_64-linux/metadata.json +4 -4
- build/torch213-cxx11-cu130-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so} +1 -1
- build/torch213-cxx11-cu130-x86_64-linux/_ops.py +3 -3
- build/torch213-cxx11-cu130-x86_64-linux/metadata.json +4 -4
- build/torch213-cxx11-cu132-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so} +1 -1
- build/torch213-cxx11-cu132-x86_64-linux/_ops.py +3 -3
- build/torch213-cxx11-cu132-x86_64-linux/metadata.json +4 -4
README.md
DELETED
|
@@ -1,9 +0,0 @@
|
|
| 1 |
-
# flashrt/fp4-gemm
|
| 2 |
-
|
| 3 |
-
This repository is a compatibility mirror for older `kernels` clients
|
| 4 |
-
that resolve repositories through the default Hugging Face model repo API.
|
| 5 |
-
|
| 6 |
-
Canonical Kernel Hub repo: https://huggingface.co/kernels/flashrt/fp4-gemm
|
| 7 |
-
|
| 8 |
-
Do not edit this mirror by hand. It is generated from the Kernel Hub
|
| 9 |
-
`vN` branches and contains the same `build/**` artifacts.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
build/torch211-cxx11-cu130-aarch64-linux/__init__.py
CHANGED
|
@@ -36,6 +36,15 @@ def _linear_fake(
|
|
| 36 |
return None
|
| 37 |
|
| 38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
@torch.library.register_fake(add_op_namespace_prefix("fp4_w4a16_linear_bf16"))
|
| 40 |
def _legacy_linear_fake(
|
| 41 |
a_packed: torch.Tensor,
|
|
@@ -54,6 +63,11 @@ def _quant_fake(x: torch.Tensor, packed: torch.Tensor, sfa: torch.Tensor, is_sfb
|
|
| 54 |
return None
|
| 55 |
|
| 56 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
@torch.library.register_fake(add_op_namespace_prefix("dequantize_fp4_sfa_fp16"))
|
| 58 |
def _dequant_fake(packed: torch.Tensor, sfa: torch.Tensor, out: torch.Tensor, is_sfb: bool = False) -> None:
|
| 59 |
return None
|
|
@@ -98,6 +112,19 @@ def quantize_fp4_sfa_fp16(
|
|
| 98 |
return packed, sfa
|
| 99 |
|
| 100 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 101 |
def dequantize_fp4_sfa_fp16(
|
| 102 |
packed: torch.Tensor,
|
| 103 |
sfa: torch.Tensor,
|
|
@@ -125,6 +152,30 @@ def nvfp4_gemm_bf16(
|
|
| 125 |
return out
|
| 126 |
|
| 127 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 128 |
def fp4_w4a16_linear_bf16(
|
| 129 |
a_packed: torch.Tensor,
|
| 130 |
b_packed: torch.Tensor,
|
|
@@ -243,6 +294,7 @@ def nvfp4_gemm_streamk_bias_bf16(
|
|
| 243 |
__all__ = [
|
| 244 |
"dequantize_fp4_sfa_fp16",
|
| 245 |
"fp4_w4a16_linear_bf16",
|
|
|
|
| 246 |
"nvfp4_gemm_bf16",
|
| 247 |
"nvfp4_gemm_bias_gelu_bf16",
|
| 248 |
"nvfp4_gemm_bias_gelu_nvfp4",
|
|
@@ -250,5 +302,6 @@ __all__ = [
|
|
| 250 |
"nvfp4_gemm_streamk_bf16",
|
| 251 |
"nvfp4_gemm_streamk_bias_bf16",
|
| 252 |
"quantize_fp4_sfa_fp16",
|
|
|
|
| 253 |
"sfa_size_bytes",
|
| 254 |
]
|
|
|
|
| 36 |
return None
|
| 37 |
|
| 38 |
|
| 39 |
+
@torch.library.register_fake(add_op_namespace_prefix("fp4_w4a4_gemv_warpsplit_bf16"))
|
| 40 |
+
def _gemv_warpsplit_fake(a_packed, b_packed, sfa, sfb, out, alpha: float = 1.0, warps: int = 4, stages: int = 4) -> None:
|
| 41 |
+
if a_packed.shape[0] != 1:
|
| 42 |
+
raise RuntimeError("warp-split GEMV serves M=1 only")
|
| 43 |
+
if out.shape != (1, b_packed.shape[0]):
|
| 44 |
+
raise RuntimeError("out must have shape (1, N)")
|
| 45 |
+
return None
|
| 46 |
+
|
| 47 |
+
|
| 48 |
@torch.library.register_fake(add_op_namespace_prefix("fp4_w4a16_linear_bf16"))
|
| 49 |
def _legacy_linear_fake(
|
| 50 |
a_packed: torch.Tensor,
|
|
|
|
| 63 |
return None
|
| 64 |
|
| 65 |
|
| 66 |
+
@torch.library.register_fake(add_op_namespace_prefix("quantize_fp4_sfa_bf16"))
|
| 67 |
+
def _quant_bf16_fake(x: torch.Tensor, packed: torch.Tensor, sfa: torch.Tensor, is_sfb: bool = False) -> None:
|
| 68 |
+
return None
|
| 69 |
+
|
| 70 |
+
|
| 71 |
@torch.library.register_fake(add_op_namespace_prefix("dequantize_fp4_sfa_fp16"))
|
| 72 |
def _dequant_fake(packed: torch.Tensor, sfa: torch.Tensor, out: torch.Tensor, is_sfb: bool = False) -> None:
|
| 73 |
return None
|
|
|
|
| 112 |
return packed, sfa
|
| 113 |
|
| 114 |
|
| 115 |
+
def quantize_fp4_sfa_bf16(
|
| 116 |
+
x: torch.Tensor,
|
| 117 |
+
packed: torch.Tensor | None = None,
|
| 118 |
+
sfa: torch.Tensor | None = None,
|
| 119 |
+
is_sfb: bool = False,
|
| 120 |
+
):
|
| 121 |
+
"""Quantize BF16 directly to packed E2M1 and CUTLASS SFA/SFB."""
|
| 122 |
+
if packed is None or sfa is None:
|
| 123 |
+
packed, sfa = _alloc_fp4(x.shape[0], x.shape[1], x.device)
|
| 124 |
+
ops.quantize_fp4_sfa_bf16(x, packed, sfa, bool(is_sfb))
|
| 125 |
+
return packed, sfa
|
| 126 |
+
|
| 127 |
+
|
| 128 |
def dequantize_fp4_sfa_fp16(
|
| 129 |
packed: torch.Tensor,
|
| 130 |
sfa: torch.Tensor,
|
|
|
|
| 152 |
return out
|
| 153 |
|
| 154 |
|
| 155 |
+
def fp4_w4a4_gemv_warpsplit_bf16(
|
| 156 |
+
a_packed: torch.Tensor,
|
| 157 |
+
b_packed: torch.Tensor,
|
| 158 |
+
sfa: torch.Tensor,
|
| 159 |
+
sfb: torch.Tensor,
|
| 160 |
+
*,
|
| 161 |
+
alpha: float = 1.0,
|
| 162 |
+
warps: int = 4,
|
| 163 |
+
stages: int = 4,
|
| 164 |
+
out: Optional[torch.Tensor] = None,
|
| 165 |
+
) -> torch.Tensor:
|
| 166 |
+
"""Warp-split-K NVFP4 W4A4 GEMV for the M=1 decode row (SM120).
|
| 167 |
+
|
| 168 |
+
Splits K across warps inside one block with a shared-memory reduce -
|
| 169 |
+
no cross-block intermediate, so it stays safe under CUDA-graph
|
| 170 |
+
replay - and fills the SMs the tiled GEMM underfills at long-K
|
| 171 |
+
small-M decode shapes. Same packed/scale layouts as the linear
|
| 172 |
+
entry points."""
|
| 173 |
+
if out is None:
|
| 174 |
+
out = torch.empty((1, b_packed.shape[0]), device=a_packed.device, dtype=torch.bfloat16)
|
| 175 |
+
ops.fp4_w4a4_gemv_warpsplit_bf16(a_packed, b_packed, sfa, sfb, out, float(alpha), int(warps), int(stages))
|
| 176 |
+
return out
|
| 177 |
+
|
| 178 |
+
|
| 179 |
def fp4_w4a16_linear_bf16(
|
| 180 |
a_packed: torch.Tensor,
|
| 181 |
b_packed: torch.Tensor,
|
|
|
|
| 294 |
__all__ = [
|
| 295 |
"dequantize_fp4_sfa_fp16",
|
| 296 |
"fp4_w4a16_linear_bf16",
|
| 297 |
+
"fp4_w4a4_gemv_warpsplit_bf16",
|
| 298 |
"nvfp4_gemm_bf16",
|
| 299 |
"nvfp4_gemm_bias_gelu_bf16",
|
| 300 |
"nvfp4_gemm_bias_gelu_nvfp4",
|
|
|
|
| 302 |
"nvfp4_gemm_streamk_bf16",
|
| 303 |
"nvfp4_gemm_streamk_bias_bf16",
|
| 304 |
"quantize_fp4_sfa_fp16",
|
| 305 |
+
"quantize_fp4_sfa_bf16",
|
| 306 |
"sfa_size_bytes",
|
| 307 |
]
|
build/torch211-cxx11-cu130-aarch64-linux/{_fp4_gemm_cuda_7781728.abi3.so → _fp4_gemm_cuda_b46a817.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49cad7b4b7a236d14238c99a832bdc106c5e8494c8ce89e000c3bc75b45cc061
|
| 3 |
+
size 751384
|
build/torch211-cxx11-cu130-aarch64-linux/_ops.py
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _fp4_gemm_cuda_b46a817
|
| 3 |
+
ops = torch.ops._fp4_gemm_cuda_b46a817
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
+
return f"_fp4_gemm_cuda_b46a817::{op_name}"
|
build/torch211-cxx11-cu130-aarch64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
@@ -13,15 +13,15 @@
|
|
| 13 |
"digest": {
|
| 14 |
"algorithm": "sha256",
|
| 15 |
"files": {
|
| 16 |
-
"__init__.py": "
|
| 17 |
-
"
|
| 18 |
-
"_ops.py": "
|
| 19 |
"fp4_gemm/__init__.py": "v6p5XMfQzddhi1fLSAw4HX9CyS0rQsidvu9VsT01xi4="
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
| 23 |
"kernel": {
|
| 24 |
-
"sha": "
|
| 25 |
"dirty": false
|
| 26 |
},
|
| 27 |
"validation": {
|
|
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
+
"id": "_fp4_gemm_cuda_b46a817",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 13 |
"digest": {
|
| 14 |
"algorithm": "sha256",
|
| 15 |
"files": {
|
| 16 |
+
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 17 |
+
"_fp4_gemm_cuda_b46a817.abi3.so": "ScrXtLeiNtFCOMmagyvcEGxehJTIzongAMO8dbRcwGE=",
|
| 18 |
+
"_ops.py": "PqcxyYzP4Co0y9NUZPOomwASQWuPPNn5QjtKrfTiZ2k=",
|
| 19 |
"fp4_gemm/__init__.py": "v6p5XMfQzddhi1fLSAw4HX9CyS0rQsidvu9VsT01xi4="
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
| 23 |
"kernel": {
|
| 24 |
+
"sha": "b46a81771fdabe26f2c2494dc2ecf8492c88c45a",
|
| 25 |
"dirty": false
|
| 26 |
},
|
| 27 |
"validation": {
|
build/torch212-cxx11-cu130-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2428552
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d34445bc1a1d2010ed592eade54e88fd2c872f135a49d18fa98d1aba9c95fe6
|
| 3 |
size 2428552
|
build/torch212-cxx11-cu130-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _fp4_gemm_cuda_0fe564a
|
| 3 |
+
ops = torch.ops._fp4_gemm_cuda_0fe564a
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_fp4_gemm_cuda_0fe564a::{op_name}"
|
build/torch212-cxx11-cu130-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
@@ -15,8 +15,8 @@
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
-
"
|
| 19 |
-
"_ops.py": "
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
@@ -26,7 +26,7 @@
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
-
"sha": "
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
+
"id": "_fp4_gemm_cuda_0fe564a",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
+
"_fp4_gemm_cuda_0fe564a.abi3.so": "bTREW8Gh0gEO1ZLq3lToj9LIcvE1pJ0Y+pjRq6nJX+Y=",
|
| 19 |
+
"_ops.py": "64wwcwE28ue+jCMSXh8vVgbUevYtei6uGZjpzaoMVrw="
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
+
"sha": "0fe564afa690cd8a731729a59f98be95ebee4b34",
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
build/torch212-cxx11-cu132-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2428504
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:88e788a3e8cc8ed5dcdc13a2bc9b846639d0e116b0f17d3f0ff882af3693c0ff
|
| 3 |
size 2428504
|
build/torch212-cxx11-cu132-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _fp4_gemm_cuda_0fe564a
|
| 3 |
+
ops = torch.ops._fp4_gemm_cuda_0fe564a
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_fp4_gemm_cuda_0fe564a::{op_name}"
|
build/torch212-cxx11-cu132-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
@@ -15,8 +15,8 @@
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
-
"
|
| 19 |
-
"_ops.py": "
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
@@ -26,7 +26,7 @@
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
-
"sha": "
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
+
"id": "_fp4_gemm_cuda_0fe564a",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
+
"_fp4_gemm_cuda_0fe564a.abi3.so": "iOeIo+jMjtXc3BOivJuEZjnQ4Raw8X0/D/iCrzaTwP8=",
|
| 19 |
+
"_ops.py": "64wwcwE28ue+jCMSXh8vVgbUevYtei6uGZjpzaoMVrw="
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
+
"sha": "0fe564afa690cd8a731729a59f98be95ebee4b34",
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
build/torch213-cxx11-cu130-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2428392
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d9cbd6af0422c5d49dd2d98ac57a58de7834902e280e1c533af843faa35700c
|
| 3 |
size 2428392
|
build/torch213-cxx11-cu130-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _fp4_gemm_cuda_0fe564a
|
| 3 |
+
ops = torch.ops._fp4_gemm_cuda_0fe564a
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_fp4_gemm_cuda_0fe564a::{op_name}"
|
build/torch213-cxx11-cu130-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
@@ -15,8 +15,8 @@
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
-
"
|
| 19 |
-
"_ops.py": "
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
@@ -26,7 +26,7 @@
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
-
"sha": "
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
+
"id": "_fp4_gemm_cuda_0fe564a",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
+
"_fp4_gemm_cuda_0fe564a.abi3.so": "LZy9avBCLF1J3S2YrFeljeeDSQLigOHFM6+EP6o1cAw=",
|
| 19 |
+
"_ops.py": "64wwcwE28ue+jCMSXh8vVgbUevYtei6uGZjpzaoMVrw="
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
+
"sha": "0fe564afa690cd8a731729a59f98be95ebee4b34",
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
build/torch213-cxx11-cu132-x86_64-linux/{_fp4_gemm_cuda_b46a817.abi3.so → _fp4_gemm_cuda_0fe564a.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2428344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b30ed92af61d4f1ac4f640a157d783e82a262857d7983f9d50aa1ebd738552e5
|
| 3 |
size 2428344
|
build/torch213-cxx11-cu132-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _fp4_gemm_cuda_0fe564a
|
| 3 |
+
ops = torch.ops._fp4_gemm_cuda_0fe564a
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_fp4_gemm_cuda_0fe564a::{op_name}"
|
build/torch213-cxx11-cu132-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
@@ -15,8 +15,8 @@
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
-
"
|
| 19 |
-
"_ops.py": "
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
@@ -26,7 +26,7 @@
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
-
"sha": "
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"name": "fp4-gemm",
|
| 3 |
+
"id": "_fp4_gemm_cuda_0fe564a",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 15 |
"algorithm": "sha256",
|
| 16 |
"files": {
|
| 17 |
"__init__.py": "pJlGLDlOKnju+m59WyZJjpynMTvkPQTF6w+z4YfCwSY=",
|
| 18 |
+
"_fp4_gemm_cuda_0fe564a.abi3.so": "sw7ZKvYdTxrE9kChV9eD6ComKFfXmD+dUKoevXOFUuU=",
|
| 19 |
+
"_ops.py": "64wwcwE28ue+jCMSXh8vVgbUevYtei6uGZjpzaoMVrw="
|
| 20 |
}
|
| 21 |
},
|
| 22 |
"provenance": {
|
|
|
|
| 26 |
"dirty": false
|
| 27 |
},
|
| 28 |
"kernel": {
|
| 29 |
+
"sha": "0fe564afa690cd8a731729a59f98be95ebee4b34",
|
| 30 |
"dirty": false
|
| 31 |
}
|
| 32 |
}
|