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 14, 05:30 UTC
[BugFix] Fix failing test quantization/test_compressed_tensors.py::test_compressed_tensors_fp8_block_enabled (#26436)
Signed-off-by: morrison-turnansky <mturnans@redhat.com>
Morrison Turnansky committed
Oct 8, 2025 at 16:04 UTC
e1ba23566891e58c3a27a62663a3095b40cb3d69
1 file changed
+1
-1
vllm/config/vllm.py
+1
-1
@@ -597,7 +597,7 @@ class VllmConfig:
597
# https://github.com/vllm-project/vllm/issues/25094
598
if has_blocked_weights():
599
custom_ops = self.compilation_config.custom_ops
600
- if "none" not in custom_ops and "-quant_fp8" not in custom_ops:
600
+ if "-quant_fp8" not in custom_ops:
601
custom_ops.append("+quant_fp8")
602
603
def update_sizes_for_sequence_parallelism(self, possible_sizes: list) -> list: