lthn/lemer
LoRA-style (fewer layers) vs LetheanNetwork/lemer. lthn/lemer matches part of LetheanNetwork/lemer (565 of its 2011 tensors; the rest of LetheanNetwork/lemer has no counterpart). On the shared part: lthn/lemer 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 35 of 35 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/2, Norms 0/247, MLP 0/105, other 0/71, Attention 100/140