saishf/Kuro-Lotus-10.7B

Full fine-tune vs BlueNipples/SnowLotus-v2-10.7B. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 1/1, Norms 97/97, MLP 144/144, Attention 188/192, lm_head 1/1

2 derivatives checked

• mradermacher/Kuro-Lotus-10.7B-i1-GGUF: Quantization
• tensorblock/Kuro-Lotus-10.7B-GGUF: Quantization

On the Hub · Downloads on Model Pulse