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 13, 14:30 UTC
[Misc] fix typo and add detailed log (#28178)
Signed-off-by: Andy Xie <andy.xning@gmail.com>
Ning Xie committed
Nov 9, 2025 at 13:33 UTC
e5e9067e61600eedd4e75bd1c512ec52872916aa
2 files changed
+2
-2
examples/offline_inference/load_sharded_state.py
+1
-1
@@ -11,7 +11,7 @@ python save_sharded_state.py \
11
--model /path/to/load \
12
--quantization deepspeedfp \
13
--tensor-parallel-size 8 \
14
- --output /path/to/save/sharded/modele
14
+ --output /path/to/save/sharded/model
15
16
python load_sharded_state.py \
17
--model /path/to/saved/sharded/model \
vllm/v1/worker/gpu_model_runner.py
+1
-1
@@ -3035,7 +3035,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
3035
time_after_load = time.perf_counter()
3036
self.model_memory_usage = m.consumed_memory
3037
logger.info_once(
3038
- "Model loading took %.4f GiB and %.6f seconds",
3038
+ "Model loading took %.4f GiB memory and %.6f seconds",
3039
self.model_memory_usage / GiB_bytes,
3040
time_after_load - time_before_load,
3041
scope="local",