aliRafik/ERNIE_45_VL_28B_A3B_PT_LaTeX_OCR_finetuned_16bit
LoRA-style vs unsloth/ERNIE-4.5-VL-28B-A3B-PT. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, fc1, fc2, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: MLP 2361/2558, Embeddings 0/2, Norms 0/188, Attention 112/240, other 0/12