TheDrummer/Rocinante-X-12B-v1
LoRA-style vs mistralai/Mistral-Nemo-Instruct-2407. 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 40 of 40 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/81, MLP 120/120, Attention 160/160, lm_head 0/1
10 derivatives checked
- bartowski/TheDrummer_Rocinante-X-12B-v1-GGUF: Same weights, converted
- mradermacher/Rocinante-X-12B-v1-i1-GGUF: Quantization
- mradermacher/Rocinante-X-12B-v1-GGUF: Quantization
- MuXodious/Rocinante-X-12B-v1-absolute-heresy: Abliterated
- ailexleon/Rocinante-X-12B-v1-mlx-6Bit: Quantization
- ailexleon/Rocinante-X-12B-v1-mlx-4Bit: Quantization
- KebalBaguette/Rocinante-X-12B-v1-AWQ: Quantization
- DavidAU/Rocinante-X-12B-v1-Heretic-Uncensored: Abliterated
- ewald1976/Red-Rosie-12B: Partial fine-tune
- ewald1976/MeterMaid-12b: Full fine-tune
1 other repos hold byte-for-byte identical weight files.