Koshkasa/ArliAI_GLM-4.5-Air-Derestricted_REAP50_Creative
LoRA-style (fewer layers) vs ArliAI/GLM-4.5-Air-Derestricted. Koshkasa/ArliAI_GLM-4.5-Air-Derestricted_REAP50_Creative matches part of ArliAI/GLM-4.5-Air-Derestricted (9195 of its 18329 tensors; the rest of ArliAI/GLM-4.5-Air-Derestricted 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, up_proj changed, in 45 of 46 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/93, MLP 1140/1297, Attention 0/322, lm_head 0/1