vicgalle/solarized-13B-dpo
LoRA-style vs vicgalle/SOLAR-13B-Instruct-v1.0. 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 56 of 56 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/113, MLP 168/168, Attention 224/224, lm_head 0/1
2 derivatives checked
- mradermacher/solarized-13B-dpo-GGUF: Quantization
- mradermacher/solarized-13B-dpo-i1-GGUF: Quantization