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 14, 05:30 UTC
[compile] Add enable_prompt_embeds to compile hash. (#27285)
Signed-off-by: zhxchen17 <zhxchen17@fb.com> Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Zhengxu Chen committed
Oct 28, 2025 at 08:46 UTC
259504e147c2818d8a2baae82c59752ab1b9a558
1 file changed
+2
-1
vllm/config/model.py
+2
-1
@@ -163,7 +163,7 @@ class ModelConfig:
163
specified by the server file system. This is a security risk. Should only
164
be enabled in trusted environments."""
165
allowed_media_domains: list[str] | None = None
166
- """If set, only media URLs that belong to this domain can be used for
166
+ """If set, only media URLs that belong to this domain can be used for
167
multi-modal inputs. """
168
revision: str | None = None
169
"""The specific model version to use. It can be a branch name, a tag name,
@@ -345,6 +345,7 @@ class ModelConfig:
345
factors.append(self.rope_scaling)
346
factors.append(self.rope_theta)
347
factors.append(self.video_pruning_rate)
348
+ factors.append(self.enable_prompt_embeds)
349
350
# hf_config can control how the model looks!
351
try: