JY623/KoSOLAR-10.7B-merge-v2.0
Full fine-tune vs chihoonlee10/T3Q-ko-solar-sft-dpo-v1.0. Fine-tune: 97% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 0/1, Norms 12/97, MLP 141/144, Attention 188/192, lm_head 1/1
3 derivatives checked
- mradermacher/KoSOLAR-10.7B-merge-v2.0-i1-GGUF: Quantization
- JY623/KoSOLRA-10.7B-merge-v2.2: LoRA-style
- JY623/KoSOLRA-10.7B-merge-v2.3: Full fine-tune