aixonlab/Zara-14b-v1.2
LoRA-style vs aixonlab/Zara-14b-v1.1. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 48 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/97, MLP 144/144, Attention 192/336, lm_head 0/1
3 derivatives checked
- mradermacher/Zara-14b-v1.2-GGUF: Quantization
- Disya/amoral-cogito-Zara-14B-ties: Full fine-tune
- Disya/amoral-cogito-Zara-14B: Full fine-tune