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, 09:15 UTC
Fix typo
Woosuk Kwon committed
Feb 23, 2023 at 20:19 UTC
86f9eb6d39f12372c7349809485d743241f58907
1 file changed
-1
cacheflow/models/opt.py
-1
@@ -179,7 +179,6 @@ class OPTDecoder(OPTPreTrainedModel):
179
) -> torch.Tensor:
180
inputs_embeds = self.embed_tokens(input_ids)
181
pos_embeds = self.embed_positions(positions)
182
- pos_embeds = None
182
if self.project_in is not None:
183
inputs_embeds = self.project_in(inputs_embeds)
184
hidden_states = inputs_embeds + pos_embeds