mradermacher/hubble-8b-100b_toks-standard-hf-GGUF

Same weights, converted vs allegrolab/hubble-8b-100b_toks-standard-hf. Same weights stored in another dtype (BF16 → F16/F32).

Tensors changed: Embeddings 0/1, Norms 0/73, MLP 0/216, Attention 0/144, lm_head 0/1

On the Hub · Downloads on Model Pulse