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 17, 05:30 UTC
[Minor] Fix err msg (#2431)
Woosuk Kwon committed
Jan 12, 2024 at 14:02 UTC
35c4bc20d9d454f58506b561b6770d3ae4752bf9
1 file changed
+3
-3
vllm/worker/worker.py
+3
-3
@@ -239,6 +239,6 @@ def _check_if_gpu_supports_dtype(torch_dtype: torch.dtype):
239
raise ValueError(
240
"Bfloat16 is only supported on GPUs with compute capability "
241
f"of at least 8.0. Your {gpu_name} GPU has compute capability "
242
- f"{compute_capability[0]}.{compute_capability[1]}."
243
- f" You can explicitly specify the data type by using the --dtype option, for example: --dtype=half."
244
- )
242
+ f"{compute_capability[0]}.{compute_capability[1]}. "
243
+ "You can use float16 instead by explicitly setting the"
244
+ "`dtype` flag in CLI, for example: --dtype=half.")