lurf21/codegemma-2b-NEP
LoRA-style vs google/codegemma-2b. 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 18 of 18 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/37, MLP 54/54, Attention 72/72
2 derivatives checked
- mradermacher/codegemma-2b-NEP-new-GGUF: Same weights, converted
- mradermacher/codegemma-2b-NES-GGUF: LoRA-style