xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-23 14:04:29 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Clean up MM profiling warnings (#25222)

Browse Source 
Signed-off-by: Roger Wang <hey@rogerw.io>
This commit is contained in:
Roger Wang 2025-09-18 21:46:57 -07:00 committed by GitHub
parent 1a0a04dae9
commit 31a8a2a7bc
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 0 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
vllm/multimodal/profiling.py
View File

@ -234,19 +234,6 @@ class MultiModalProfiler(Generic[_I]):
prompt_token_ids = mm_inputs["prompt_token_ids"]
total_len = len(prompt_token_ids)
# V0 does not support chunked prefill.
if total_len > seq_len and not envs.VLLM_USE_V1:
# `max_num_batched_tokens` is defined by `SchedulerConfig`
logger.warning_once(
"The sequence length used for profiling (max_num_batched_tokens / max_num_seqs = %d) " # noqa: E501
"is too short to hold the multi-modal embeddings in the worst case (%d tokens in total, out of which %s are reserved for multi-modal embeddings). " # noqa: E501
"This may cause certain multi-modal inputs to fail during inference, even when the input text is short. " # noqa: E501
"To avoid this, you should increase `max_model_len`, reduce `max_num_seqs`, and/or reduce `mm_counts`.", # noqa: E501
seq_len,
total_len,
str(self._get_mm_num_tokens(mm_inputs)),
)
if total_len < seq_len:
prompt_token_ids.extend([0] * (seq_len - total_len))
@ -270,22 +257,6 @@ class MultiModalProfiler(Generic[_I]):
mm_counts=mm_counts,
)
if max_tokens_per_item is not None:
if mm_counts is None:
total_mm_tokens = sum(max_tokens_per_item.values())
else:
total_mm_tokens = sum(max_tokens_per_item[k] * mm_counts[k]
for k in max_tokens_per_item.keys()
& mm_counts.keys())
if total_mm_tokens > seq_len:
logger.warning_once(
"The sequence length (%d) is smaller than the pre-defined"
" worst-case total number of multimodal tokens (%d). "
"This may cause certain multi-modal inputs to fail during "
"inference. To avoid this, you should increase "
"`max_model_len` or reduce `mm_counts`.",
seq_len,
total_mm_tokens,
)
return max_tokens_per_item
mm_inputs = self._get_dummy_mm_inputs(seq_len, mm_counts)