Skip to content

vllm.model_executor.models.musicflamingo

MusicFlamingo model adapter.

MusicFlamingo shares the AudioFlamingo3 architecture, so we reuse the same implementation and multimodal processor, while accepting MusicFlamingo config and processor classes when available.

MusicFlamingoForConditionalGeneration

Bases: AudioFlamingo3ForConditionalGeneration

MusicFlamingo model for conditional generation.

Source code in vllm/model_executor/models/musicflamingo.py 
64
65
66
67
68
69
70
@MULTIMODAL_REGISTRY.register_processor(
    AudioFlamingo3MultiModalProcessor,
    info=MusicFlamingoProcessingInfo,
    dummy_inputs=MusicFlamingoDummyInputsBuilder,
)
class MusicFlamingoForConditionalGeneration(AudioFlamingo3ForConditionalGeneration):
    """MusicFlamingo model for conditional generation."""