prithivMLmods/Gelato-30B-A3B-f32-AIO-GGUF

Same weights, converted vs mlfoundations/Gelato-30B-A3B. Same weights stored in another dtype (BF16 → F16/F32), plus 144 new tensors.

Tensors changed: Attention 0/192, MLP 0/48, Norms 0/193, Embeddings 0/1, lm_head 0/1

On the Hub · Downloads on Model Pulse