prithivMLmods/Hoags-2B-Exp
LoRA-style vs prithivMLmods/Qwen2-VL-OCR-2B-Instruct. 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, proj, q_proj, qkv, up_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/2, Norms 0/187, MLP 148/216, Attention 176/324
2 derivatives checked
- mradermacher/Hoags-2B-Exp-GGUF: Same weights, converted (fewer layers)
- csabakecskemeti/prithivMLmods.Hoags-2B-Exp-GGUF: Same weights, converted (fewer layers)