FINAL-Bench/Darwin-9B-NEG-x-Negentropy-V8
LoRA-style (fewer layers) vs FINAL-Bench/Darwin-9B-NEG. FINAL-Bench/Darwin-9B-NEG-x-Negentropy-V8 matches part of FINAL-Bench/Darwin-9B-NEG (427 of its 775 tensors; the rest of FINAL-Bench/Darwin-9B-NEG has no counterpart). On the shared part: Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, up_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, MLP 96/96, Attention 0/224, Norms 0/105
2 derivatives checked
- mradermacher/Darwin-9B-NEG-x-Negentropy-V8-i1-GGUF: Quantization
- mradermacher/Darwin-9B-NEG-x-Negentropy-V8-GGUF: Same weights, converted