Delta-Vector/Hamanasu-4B-Instruct-KTO-V1
Full fine-tune vs Delta-Vector/Hamanasu-4B-Instruct. Fine-tune: 91% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 0/1, Norms 13/65, MLP 96/96, Attention 128/128, lm_head 1/1
3 derivatives checked
- mradermacher/Hamanasu-4B-Instruct-KTO-V1-i1-GGUF: Quantization
- mradermacher/Hamanasu-4B-Instruct-KTO-V1-GGUF: Same weights, converted
- Delta-Vector/Hamanasu-4B-Instruct-KTO-V2: Full fine-tune
1 other repos hold byte-for-byte identical weight files.