athirdpath/LD-Zephyria-37b_HEALED
LoRA-style vs SteelStorage/LD-Zephyria-37b. 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 95 of 95 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/191, MLP 285/285, Attention 380/380
3 derivatives checked
- mradermacher/LD-Zephyria-37b_HEALED-i1-GGUF: Quantization
- mradermacher/LD-Zephyria-37b_HEALED-GGUF: Quantization
- athirdpath/Cosmos-37b-BASE: Full fine-tune