Export Marlin and quantization utilities
Browse files
torch-ext/quantization/__init__.py
CHANGED
|
@@ -19,6 +19,11 @@ from .scalar_type import (
|
|
| 19 |
)
|
| 20 |
from ._ops import ops
|
| 21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
__all__ = [
|
| 24 |
"ScalarType",
|
|
@@ -32,7 +37,11 @@ __all__ = [
|
|
| 32 |
"gptq_marlin_repack",
|
| 33 |
"marlin_gemm",
|
| 34 |
"marlin_qqq_gemm",
|
|
|
|
|
|
|
|
|
|
| 35 |
"ops",
|
|
|
|
| 36 |
"scalar_types",
|
| 37 |
"scaled_fp8_quant",
|
| 38 |
"scaled_int8_quant",
|
|
|
|
| 19 |
)
|
| 20 |
from ._ops import ops
|
| 21 |
|
| 22 |
+
from .utils import marlin_utils
|
| 23 |
+
from .utils import marlin_utils_fp4
|
| 24 |
+
from .utils import marlin_utils_fp8
|
| 25 |
+
from .utils import quant_utils
|
| 26 |
+
|
| 27 |
|
| 28 |
__all__ = [
|
| 29 |
"ScalarType",
|
|
|
|
| 37 |
"gptq_marlin_repack",
|
| 38 |
"marlin_gemm",
|
| 39 |
"marlin_qqq_gemm",
|
| 40 |
+
"marlin_utils",
|
| 41 |
+
"marlin_utils_fp4",
|
| 42 |
+
"marlin_utils_fp8",
|
| 43 |
"ops",
|
| 44 |
+
"quant_utils",
|
| 45 |
"scalar_types",
|
| 46 |
"scaled_fp8_quant",
|
| 47 |
"scaled_int8_quant",
|