usermma/Huihui-LFM2.5-8B-A1B-abliterated-mlx-fp16

Same weights, converted vs huihui-ai/Huihui-LFM2.5-8B-A1B-abliterated. Same weights stored in another dtype (BF16/F32 → F16/F32), plus 72 new tensors.

Tensors changed: MLP 0/44, Embeddings 0/2, other 0/36, Norms 0/60, Attention 0/24

On the Hub · Downloads on Model Pulse