Uploaded using `kernel-builder`.
Browse files
README.md
CHANGED
|
@@ -11,7 +11,7 @@ This is the repository card of kernels-community/finegrained-fp8 that has been p
|
|
| 11 |
# make sure `kernels` is installed: `pip install -U kernels`
|
| 12 |
from kernels import get_kernel
|
| 13 |
|
| 14 |
-
kernel_module = get_kernel("kernels-community/finegrained-fp8", version=
|
| 15 |
fp8_act_quant = kernel_module.fp8_act_quant
|
| 16 |
|
| 17 |
fp8_act_quant(...)
|
|
@@ -19,19 +19,9 @@ fp8_act_quant(...)
|
|
| 19 |
|
| 20 |
## Available functions
|
| 21 |
- `fp8_act_quant`
|
| 22 |
-
- `
|
| 23 |
-
- `w4a8_block_dynamic_fp4_matmul`
|
| 24 |
-
- `w8a8_block_dynamic_fp8_matmul`
|
| 25 |
-
- `w8a8_block_static_fp8_matmul`
|
| 26 |
-
- `w8a8_tensor_dynamic_fp8_matmul`
|
| 27 |
- `matmul_batched`
|
| 28 |
-
- `w4a8_block_dynamic_fp4_matmul_batched`
|
| 29 |
-
- `w8a8_block_dynamic_fp8_matmul_batched`
|
| 30 |
-
- `w8a8_tensor_dynamic_fp8_matmul_batched`
|
| 31 |
- `matmul_grouped`
|
| 32 |
-
- `w4a8_block_dynamic_fp4_matmul_grouped`
|
| 33 |
-
- `w8a8_block_dynamic_fp8_matmul_grouped`
|
| 34 |
-
- `w8a8_tensor_dynamic_fp8_matmul_grouped`
|
| 35 |
|
| 36 |
## Benchmarks
|
| 37 |
|
|
|
|
| 11 |
# make sure `kernels` is installed: `pip install -U kernels`
|
| 12 |
from kernels import get_kernel
|
| 13 |
|
| 14 |
+
kernel_module = get_kernel("kernels-community/finegrained-fp8", version=3)
|
| 15 |
fp8_act_quant = kernel_module.fp8_act_quant
|
| 16 |
|
| 17 |
fp8_act_quant(...)
|
|
|
|
| 19 |
|
| 20 |
## Available functions
|
| 21 |
- `fp8_act_quant`
|
| 22 |
+
- `matmul_2d`
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
- `matmul_batched`
|
|
|
|
|
|
|
|
|
|
| 24 |
- `matmul_grouped`
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
## Benchmarks
|
| 27 |
|