[VLM] Remove BaseProcessingInfo.get_mm_max_tokens_per_item (#16408)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-04-10 09:06:58 -07:00
committed by GitHub
parent 7678fcd5b6
commit 83b824c8b4
39 changed files with 104 additions and 677 deletions
-12
View File
@@ -2,7 +2,6 @@
# Adapted from https://github.com/fixie-ai/ultravox/blob/ecd58c4041030bae2ad15aa6bcf04ab43199ea02/ultravox/model/ultravox_model.py
"""PyTorch Ultravox model."""
import math
from collections.abc import Iterable, Mapping, Sequence
from functools import cached_property
from typing import Any, Literal, Optional, Set, Tuple, TypedDict, Union
@@ -107,17 +106,6 @@ class UltravoxProcessingInfo(BaseProcessingInfo):
def get_supported_mm_limits(self) -> Mapping[str, Optional[int]]:
return {"audio": None}
def get_mm_max_tokens_per_item(
self,
seq_len: int,
mm_counts: Mapping[str, int],
) -> Mapping[str, int]:
feature_extractor = self.get_feature_extractor()
max_audio_tokens = math.ceil(feature_extractor.chunk_length *
_AUDIO_TOKENS_PER_SECOND)
return {"audio": max_audio_tokens * _MAX_ENCODER_BATCH_SIZE}
class UltravoxDummyInputsBuilder(BaseDummyInputsBuilder[UltravoxProcessingInfo]
):