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] Convert untraceable GroupShape to list for AMD impl (#26535)
Signed-off-by: Lucas Kabela <lucaskabela@meta.com>
Lucas Kabela committed
Oct 10, 2025 at 06:32 UTC
213b64452a4afe5ed0a38fb7cff4076705ec84e1
1 file changed
+2
-2
vllm/model_executor/layers/quantization/utils/fp8_utils.py
+2
-2
@@ -360,7 +360,7 @@ class W8A8BlockFp8LinearOp:
360
weight,
361
input_scale,
362
weight_scale,
363
- self.weight_group_shape,
363
+ list(self.weight_group_shape),
364
input_2d.dtype,
365
)
366
@@ -377,7 +377,7 @@ class W8A8BlockFp8LinearOp:
377
weight,
378
input_scale,
379
weight_scale,
380
- self.weight_group_shape,
380
+ list(self.weight_group_shape),
381
input_2d.dtype,
382
)
383