mradermacher/k-4b-GGUF
LoRA-style (fewer layers) vs ThomasTheMaker/k-4b. mradermacher/k-4b-GGUF matches part of ThomasTheMaker/k-4b (444 of its 883 tensors; the rest of ThomasTheMaker/k-4b has no counterpart). On the shared part: mradermacher/k-4b-GGUF is a GGUF (F16) of different weights. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 34 of 34 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/205, MLP 102/102, Attention 136/136