bartowski/baidu_ERNIE-4.5-21B-A3B-PT-GGUF
LoRA-style vs baidu/ERNIE-4.5-21B-A3B-PT. bartowski/baidu_ERNIE-4.5-21B-A3B-PT-GGUF is a GGUF (BF16) of different weights. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of gate changed, in 27 of 28 layers; norms, biases and embeddings are untouched.
Tensors changed: MLP 27/111, Embeddings 0/1, Norms 0/57, Attention 0/112