matchaaaaa/Chaifighter-20B-v2
Partial fine-tune (fewer layers) vs Sao10K/Fimbulvetr-11B-v2. Sao10K/Fimbulvetr-11B-v2 matches part of matchaaaaa/Chaifighter-20B-v2 (435 of its 795 tensors; the rest of matchaaaaa/Chaifighter-20B-v2 has no counterpart). On the shared part: Partial change: 17% of the weights changed (attention, lm_head, mlp, norms), in 8 of 48 layers.
Tensors changed: Embeddings 0/1, Norms 17/97, MLP 24/144, Attention 32/192, lm_head 1/1
3 derivatives checked
- mradermacher/Chaifighter-20B-v2-i1-GGUF: Quantization
- mradermacher/Chaifighter-20B-v2-GGUF: Quantization
- backyardai/Chaifighter-20B-v2-GGUF: Same weights, converted