suayptalha/Sungur-9B
LoRA-style vs ytu-ce-cosmos/Turkish-Gemma-9b-v0.1. 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 42 of 42 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/169, MLP 126/126, Attention 168/168
3 derivatives checked
- suayptalha/Sungur-9B-GGUF: Same weights, converted
- mradermacher/Sungur-9B-i1-GGUF: Quantization
- mradermacher/Sungur-9B-GGUF: Same weights, converted