xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-01 19:27:08 +08:00
Code Issues Packages Projects Releases Wiki Activity

optional input scales

Browse Source 
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
This commit is contained in:
vllmellm 2025-11-12 07:43:16 +00:00
parent 4c596a0b1e
commit 65ecf487ad
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tests/utils.py
View File

@ -1325,7 +1325,8 @@ class TestFP8Layer(torch.nn.Module):
weight_quant_key (QuantKey): Key for weight quantization configuration. weight_quant_key (QuantKey): Key for weight quantization configuration.
weight (torch.Tensor): Weight tensor for linear transformation. weight (torch.Tensor): Weight tensor for linear transformation.
weight_scale (torch.Tensor): Per-tensor or per-group scale for weights. weight_scale (torch.Tensor): Per-tensor or per-group scale for weights.
input_scale (torch.Tensor): Scale tensor for input quantization. input_scale (torch.Tensor, optional): Scale tensor for input quantization.
Defaults to None.
out_dtype (torch.dtype, optional): Output tensor data type. out_dtype (torch.dtype, optional): Output tensor data type.
Defaults to torch.get_default_dtype(). Defaults to torch.get_default_dtype().
""" """
@ -1336,7 +1337,7 @@ class TestFP8Layer(torch.nn.Module):
weight_quant_key: QuantKey, weight_quant_key: QuantKey,
weight: torch.Tensor, weight: torch.Tensor,
weight_scale: torch.Tensor, weight_scale: torch.Tensor,
input_scale: torch.Tensor, input_scale: torch.Tensor | None = None,
out_dtype: torch.dtype | None = None, out_dtype: torch.dtype | None = None,
force_kernel: FP8ScaledMMLinearKernel | None = None, force_kernel: FP8ScaledMMLinearKernel | None = None,
): ):