mlx-community/diar_sortformer_4spk-v1-fp16
Same weights, converted vs nvidia/diar_sortformer_4spk-v1. Same weights stored in another dtype (F32/I64 → F16/I64).
Tensors changed: Norms 0/342, other 0/128, MLP 0/216, Attention 0/324, Embeddings 0/1
Same weights, converted vs nvidia/diar_sortformer_4spk-v1. Same weights stored in another dtype (F32/I64 → F16/I64).
Tensors changed: Norms 0/342, other 0/128, MLP 0/216, Attention 0/324, Embeddings 0/1