xinyun/DeepSeek-V3
1
0
Fork 0
mirror of https://git.datalinker.icu/deepseek-ai/DeepSeek-V3.git synced 2026-03-16 14:27:26 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge 37f0b41ac6501266a661ebda3b1eff7bea9f3a1b into 9b4e9788e4a3a731f7567338ed15d3ec549ce03b

Browse Source
This commit is contained in:
Arupm 2025-08-29 20:56:48 +03:00 committed by GitHub
commit 79571aa67c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
inference/kernel.py
View File

@ -37,7 +37,7 @@ def act_quant_kernel(x_ptr, y_ptr, s_ptr, BLOCK_SIZE: tl.constexpr, scale_fmt: t
def act_quant(x: torch.Tensor, block_size: int = 128, scale_fmt: Optional[str] = None) -> Tuple[torch.Tensor, torch.Tensor]: def act_quant(x: torch.Tensor, block_size: int = 128, scale_fmt: Optional[str] = None) -> Tuple[torch.Tensor, torch.Tensor]:
""" """
Quantizes the input tensor `x` using block-wise quantization. Quantizes the input tensor `x` using block-wise quantization as implemented in the given function.
Args: Args:
x (torch.Tensor): The input tensor to be quantized. Must be contiguous and its last dimension size must be divisible by `block_size`. x (torch.Tensor): The input tensor to be quantized. Must be contiguous and its last dimension size must be divisible by `block_size`.