A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 12:15 UTC
[Bugfix] Disable gptq_bitblas for <SM80 to fix GPTQ on V100/T4 (#17541)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
May 1, 2025 at 18:59 UTC
24aebae1777288503657e4163e14d854ed4ab633
1 file changed
+1
-1
vllm/model_executor/layers/quantization/gptq_bitblas.py
+1
-1
@@ -134,7 +134,7 @@ class GPTQBitBLASConfig(QuantizationConfig):
134
135
@classmethod
136
def get_min_capability(cls) -> int:
137
- return 70
137
+ return 80
138
139
@classmethod
140
def get_config_filenames(cls) -> List[str]: