lthn/lemma
LoRA-style (fewer layers) vs LetheanNetwork/lemma. lthn/lemma matches part of LetheanNetwork/lemma (677 of its 2130 tensors; the rest of LetheanNetwork/lemma has no counterpart). On the shared part: lthn/lemma is a GGUF (BF16) of different weights. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, o_proj, q_proj, v_proj changed, in 42 of 42 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/2, Norms 0/296, MLP 0/126, other 0/85, Attention 132/168