Netuoso/qwepus
LoRA-style (fewer layers) vs empero-ai/Qwythos-9B-v2. Netuoso/qwepus matches part of empero-ai/Qwythos-9B-v2 (427 of its 775 tensors; the rest of empero-ai/Qwythos-9B-v2 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, in_proj_a, in_proj_b, in_proj_qkv, in_proj_z, k_proj, o_proj, out_proj, q_proj, up_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/105, Attention 152/224, MLP 96/96
2 derivatives checked
- mradermacher/qwepus-i1-GGUF: Quantization
- mradermacher/qwepus-GGUF: Same weights, converted