|
1 |
| -from .nf4tensor import NF4Tensor, to_nf4 |
2 |
| - |
3 |
| -# from ..prototype.dtypes.uint2 import UInt2Tensor, BitnetTensor |
4 |
| -from .uintx import UInt4Tensor |
5 | 1 | from .affine_quantized_tensor import (
|
6 | 2 | AffineQuantizedTensor,
|
7 |
| - to_affine_quantized_intx, |
8 |
| - to_affine_quantized_intx_static, |
9 |
| - # experimental, will be merged into floatx in the future |
10 |
| - to_affine_quantized_fpx, |
11 |
| - to_affine_quantized_floatx, |
12 |
| - to_affine_quantized_floatx_static, |
13 |
| -) |
14 |
| - |
15 |
| -from . import affine_quantized_tensor_ops |
16 |
| -from .utils import ( |
| 3 | + Float8AQTTensorImpl, |
| 4 | + Float8Layout, |
17 | 5 | Layout,
|
| 6 | + MarlinSparseLayout, |
18 | 7 | PlainLayout,
|
19 | 8 | )
|
20 | 9 | from .floatx import (
|
|
31 | 20 | UInt4Tensor,
|
32 | 21 | SemiSparseLayout,
|
33 | 22 | TensorCoreTiledLayout,
|
34 |
| - MarlinSparseLayout, |
35 |
| - PlainAQTTensorImpl, |
36 |
| - BlockSparseLayout, |
| 23 | + to_affine_quantized_floatx, |
| 24 | + to_affine_quantized_floatx_static, |
| 25 | + # experimental, will be merged into floatx in the future |
| 26 | + to_affine_quantized_fpx, |
| 27 | + to_affine_quantized_intx, |
| 28 | + to_affine_quantized_intx_static, |
37 | 29 | )
|
| 30 | +from .nf4tensor import NF4Tensor, to_nf4 |
| 31 | + |
| 32 | +# from ..prototype.dtypes.uint2 import UInt2Tensor, BitnetTensor |
| 33 | +from .uint4 import UInt4Tensor |
38 | 34 |
|
39 | 35 | __all__ = [
|
40 | 36 | "NF4Tensor",
|
|
0 commit comments