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

Make name of compressed-tensors quant method consistent across vLLM (#17255)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-04-28 17:28:13 +01:00
committed by GitHub
parent f94886946e
commit b6dd32aa07
5 changed files with 10 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
vllm/config.py
View File

@@ -752,9 +752,8 @@ class ModelConfig:
supported_quantization = QUANTIZATION_METHODS
optimized_quantization_methods = [
"fp8", "marlin", "modelopt", "gptq_marlin_24", "gptq_marlin",
"awq_marlin", "fbgemm_fp8", "compressed_tensors",
"compressed-tensors", "experts_int8", "quark", "nvfp4", "bitblas",
"gptq_bitblas"
"awq_marlin", "fbgemm_fp8", "compressed-tensors", "experts_int8",
"quark", "nvfp4", "bitblas", "gptq_bitblas"
]
if self.quantization is not None:
self.quantization = self.quantization.lower()
@@ -764,6 +763,9 @@ class ModelConfig:
if quant_cfg is not None:
quant_method = quant_cfg.get("quant_method", "").lower()
quant_method = quant_method.replace("compressed_tensors",
"compressed-tensors")
quant_cfg["quant_method"] = quant_method
# Detect which checkpoint is it
for name in QUANTIZATION_METHODS: