torch
torchaudio
numpy
# Default for Thinker replay. The B1 local extractor also supports 4.57.6;
# keep that version in existing GPU containers (see docs/local_feature_baseline.md).
transformers==4.52.3
accelerate
qwen-omni-utils[decord]
librosa
Pillow
tqdm
scikit-learn
PyYAML
einops

# Used by train_omni_turn_lora.py; already included in the ms-swift image.
peft

# Optional on compatible NVIDIA GPUs: flash-attn
# Optional CPU video backend: torchcodec
# Optional NVDEC backend: PyNvVideoCodec