mradermacher/GigaChat3-5B-pruned-GGUF

Same weights, converted vs Griniy/GigaChat3-5B-pruned. Same weights stored in another dtype (BF16 → F16/F32), plus 152 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/79, MLP 0/103, Attention 0/78, lm_head 0/1

On the Hub · Downloads on Model Pulse