heyongxin233/DETree
LoRA-style vs FacebookAI/roberta-large. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of query, value changed, in 24 of 24 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/96, Attention 48/192, other 0/96, Embeddings 0/5