mlx-community/MOSS-Audio-Tokenizer-v2-bf16

Same weights, converted vs OpenMOSS-Team/MOSS-Audio-Tokenizer-v2. Same weights stored in another dtype (F32 → BF16).

Tensors changed: other 0/622, MLP 0/368, Norms 0/736, Attention 0/368

On the Hub · Downloads on Model Pulse