rcmorano/protoLabsAI-Agents-A1-MTP-ROCMFPX
LoRA-style vs protoLabsAI/Agents-A1-MTP-GGUF. rcmorano/protoLabsAI-Agents-A1-MTP-ROCMFPX is a GGUF (Q4_0_ROCMFP4) of different weights. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of attn_output, attn_qkv, attn_v, ffn_down_exps, ffn_down_shexp, ffn_gate_exps, ffn_gate_shexp changed, in 0 of 0 layers; norms, biases and embeddings are untouched.
Tensors changed: MLP 124/165, lm_head 0/1, Norms 0/138, other 0/91, Attention 52/52