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 16, 23:15 UTC
[Bugfix] fix lora_dtype value type in arg_utils.py (#5398)
Ali Panahi committed
Jun 11, 2024 at 10:40 UTC
00e6a2dc535c89ac7c92551ef9b92acd8664df02
1 file changed
+1
-1
vllm/engine/arg_utils.py
+1
-1
@@ -68,7 +68,7 @@ class EngineArgs:
68
fully_sharded_loras: bool = False
69
lora_extra_vocab_size: int = 256
70
long_lora_scaling_factors: Optional[Tuple[float]] = None
71
- lora_dtype = 'auto'
71
+ lora_dtype: str = 'auto'
72
max_cpu_loras: Optional[int] = None
73
device: str = 'auto'
74
ray_workers_use_nsight: bool = False