TheBigBlender/Daisuke
Full fine-tune (fewer layers) vs PygmalionAI/pygmalion-2-13b. PygmalionAI/pygmalion-2-13b matches part of TheBigBlender/Daisuke (363 of its 561 tensors; the rest of TheBigBlender/Daisuke has no counterpart). On the shared part: Fine-tune: 99% of the weights changed, across 40 of 40 layers.
Tensors changed: Embeddings 1/1, Norms 80/81, MLP 120/120, Attention 160/160, lm_head 0/1
1 derivatives checked
- mradermacher/Daisuke-GGUF: Quantization