mradermacher/Trinity-Nano-Base-GGUF

Same weights, converted vs arcee-ai/Trinity-Nano-Base. Same weights stored in another dtype (BF16 → F16/F32), plus 216 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/337, MLP 0/222, Attention 0/280, lm_head 0/1

On the Hub · Downloads on Model Pulse