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, 05:45 UTC
[V1] Enable profile for LLMEngine (#10665)
Kunshang Ji committed
Nov 26, 2024 at 18:36 UTC
1f6584ee851501cfae672973b9e55d000729818c
1 file changed
+3
-3
vllm/v1/engine/llm_engine.py
+3
-3
@@ -161,13 +161,13 @@ class LLMEngine:
161
# TODO(rob): Can we get rid of these?
162
163
def get_model_config(self):
164
- pass
164
+ return self.model_config
165
166
def start_profile(self):
167
- pass
167
+ self.engine_core.profile(True)
168
169
def stop_profile(self):
170
- pass
170
+ self.engine_core.profile(False)
171
172
def get_tokenizer_group(self, group_type):
173
pass