biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Docs] Fix warnings in mkdocs build (continued) (#24092)

Browse Source 
Signed-off-by: Zerohertz <ohg3417@gmail.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
This commit is contained in:
Hyogeun Oh (오효근)
2025-09-10 22:23:28 +09:00
committed by GitHub
parent c0bd6a684a
commit ccee371e86
10 changed files with 337 additions and 342 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
vllm/model_executor/layers/quantization/bitblas.py
View File

@@ -202,7 +202,7 @@ class BitBLASLinearMethod(LinearMethodBase):
output_size: int,
params_dtype: torch.dtype,
**extra_weight_attrs,
):
) -> None:
"""Creates quantized weights for use in linear operations.
The function initializes and returns a dictionary containing quantized
@@ -211,7 +211,7 @@ class BitBLASLinearMethod(LinearMethodBase):
Args:
input_size_per_partition: The size of the input partition.
output_size_per_partition: The size of the output partition.
output_partition_sizes: List of output partition sizes.
input_size: The total size of the input (unused).
output_size: The total size of the output (unused).
params_dtype:
@@ -222,9 +222,9 @@ class BitBLASLinearMethod(LinearMethodBase):
scales ('scales'), and zeros ('zeros').
Raises:
ValueError: If `params_dtype` is not `torch.float16` or if the
input size per partition is not divisible by the group size in
`quant_config`.
ValueError: If `params_dtype` is not `torch.float16` or if the input
size per partition is not divisible by the group size
in `quant_config`.
"""
del input_size, output_size # Unused arguments.
weight_loader = extra_weight_attrs["weight_loader"]

6
vllm/model_executor/layers/quantization/gptq_bitblas.py
View File

@@ -265,9 +265,9 @@ class GPTQBitBLASLinearMethod(LinearMethodBase):
scales ('scales'), and zeros ('zeros').
Raises:
ValueError: If `params_dtype` is not `torch.float16` or
if the input size per partition is not divisible by the
group size in `quant_config`.
ValueError: If `params_dtype` is not `torch.float16` or if the input
size per partition is not divisible by the group size
in `quant_config`.
"""
if params_dtype != torch.float16:
raise ValueError("Parameter data type must be torch.float16, "

8
vllm/model_executor/layers/quantization/kernels/mixed_precision/__init__.py
View File

@@ -46,11 +46,11 @@ def choose_mp_linear_kernel(
performance.
Args:
config (MPLinearLayerConfig): Description of the linear layer to be
implemented.
config (MPLinearLayerConfig): Description of the linear layer to be
implemented.
compute_capability (Optional[int], optional): The compute capability of
the target device, if None uses `current_platform` to get the compute
capability. Defaults to None.
the target device, if None uses `current_platform` to get
the compute capability. Defaults to None.
Raises:
ValueError: If no kernel can implement the given config.