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

[CI/Build] upgrade compressed-tensors to 0.12.2 to address LGPLv3 (#26501)

Browse Source 
Signed-off-by: Sangyeon Cho <josang1204@gmail.com>
This commit is contained in:
Sangyeon Cho
2025-10-14 01:58:33 +09:00
committed by GitHub
parent 5c7fe25491
commit a1b2d658ee
3 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors.py
View File

@@ -310,7 +310,7 @@ class CompressedTensorsConfig(QuantizationConfig):
)
is_float_type = (
weight_quant.type == QuantizationType.FLOAT
and input_quant.type == QuantizationType.FLOAT.value
and input_quant.type == QuantizationType.FLOAT
)
is_4_bits = weight_quant.num_bits == 4 and input_quant.num_bits == 4

2
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py
View File

@@ -143,7 +143,7 @@ class CompressedTensorsMoEMethod(FusedMoEMethodBase):
# Prefer to use the MarlinMoE kernel when it is supported.
if not check_moe_marlin_supports_layer(layer, group_size):
if (
weight_quant.strategy in QuantizationStrategy.GROUP
weight_quant.strategy == QuantizationStrategy.GROUP
and weight_quant.actorder
in (ActivationOrdering.GROUP, ActivationOrdering.DYNAMIC)
):