mradermacher/InnoSpark3.0-9B-260630-GGUF

Same weights, converted vs sii-research/InnoSpark3.0-9B-260630. Same weights stored in another dtype (BF16 → F16/F32), plus 39 new tensors.

Tensors changed: MLP 0/96, Norms 0/81, Attention 0/32, lm_head 0/1, Embeddings 0/1

On the Hub · Downloads on Model Pulse