Diamegs/PIT-4B-FT-202112
LoRA-style vs Diamegs/PIT-4B-202112. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of c_fc, c_k, c_proj, c_q, c_v changed, in 20 of 20 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Attention 80/80, MLP 40/40, Embeddings 0/1
1 other repos hold byte-for-byte identical weight files.