Delta-Vector/Pascal-12B
LoRA-style vs Delta-Vector/Pascal-Base-12B. 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 48 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, Norms 0/400, MLP 144/252, Attention 192/408, other 0/1
5 derivatives checked
- mradermacher/Gemma-LN-SFT-GGUF: Quantization (fewer layers)
- mradermacher/Pascal-12B-GGUF: Quantization (fewer layers)
- mradermacher/Pascal-12B-i1-GGUF: Quantization (fewer layers)
- Delta-Vector/Daichi-12B: LoRA-style
- Edens-Gate/Honeyfeed-LN-merge: LoRA-style