ai-sage/GigaChat-20B-A3B-instruct
Full fine-tune vs ai-sage/GigaChat-20B-A3B-base. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 57/57, MLP 2829/2829, Attention 112/112, lm_head 1/1
4 derivatives checked
- ai-sage/GigaChat-20B-A3B-instruct-GGUF: Same weights, converted
- mradermacher/GigaChat-20B-A3B-instruct-i1-GGUF: Quantization
- mradermacher/GigaChat-20B-A3B-instruct-GGUF: Quantization
- ai-sage/GigaChat-20B-A3B-instruct-bf16: Same weights, converted