Laxhar/noobai-XL-Vpred-0.65
Partial fine-tune vs Laxhar/noobai-XL-Vpred-0.6. Partial change: 72% of the weights changed (attention, embeddings, norms, other), in 0 of 12 layers.
Tensors changed: other 534/664, Attention 700/716, Norms 442/546, Embeddings 4/590
1 derivatives checked
- Laxhar/noobai-XL-Vpred-0.75: Partial fine-tune