AMAImedia/NOESIS-Qwopus3.5-9B-Coder-v3.5-BF16
LoRA-style (fewer layers) vs Jackrong/Qwopus3.5-9B-v3.5. AMAImedia/NOESIS-Qwopus3.5-9B-Coder-v3.5-BF16 matches part of Jackrong/Qwopus3.5-9B-v3.5 (427 of its 775 tensors; the rest of Jackrong/Qwopus3.5-9B-v3.5 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, 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, MLP 96/96, Attention 56/224, Norms 0/105