biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[UX] Improve error message for MM input too long (#39409)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-04-09 21:20:19 +08:00
committed by GitHub
parent 6c749399b7
commit 54d244f28f
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/v1/engine/input_processor.py
View File

@@ -405,11 +405,11 @@ class InputProcessor:
decoder_mm_positions = prompt_input["mm_placeholders"]
for modality, mm_positions in decoder_mm_positions.items():
for mm_position in mm_positions:
embed_length = mm_position.get_num_embeds()
if embed_length > self.mm_encoder_cache_size:
num_embeds = mm_position.get_num_embeds()
if num_embeds > self.mm_encoder_cache_size:
raise ValueError(
f"The {prompt_type} prompt contains a(n) {modality} item "
f"with length {embed_length}, which exceeds the "
f"with {num_embeds} embedding tokens, which exceeds the "
f"pre-allocated encoder cache size "
f"{self.mm_encoder_cache_size}. Please reduce the input "
f"size or increase the encoder cache size "