mrcuddle/Lumimaid-v0.2-12B-Pixtral
Full fine-tune (fewer layers) vs NeverSleep/Lumimaid-v0.2-12B. NeverSleep/Lumimaid-v0.2-12B matches part of mrcuddle/Lumimaid-v0.2-12B-Pixtral (361 of its 585 tensors; the rest of mrcuddle/Lumimaid-v0.2-12B-Pixtral has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 40 of 40 layers.
Tensors changed: Norms 81/81, MLP 120/120, Attention 160/160
1 derivatives checked
- Koitenshin/Lumimaid_VISION-v0.2-12B-Pixtral-GGUF: Quantization (fewer layers)
13 other repos hold byte-for-byte identical weight files.