rinna/nekomata-7b

Full fine-tune vs Qwen/Qwen-7B. Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Attention 96/96, Norms 65/65, MLP 96/96, Embeddings 1/1, lm_head 1/1

3 derivatives checked

• mradermacher/nekomata-7b-i1-GGUF: Quantization
• mradermacher/nekomata-7b-GGUF: Same weights, converted
• rinna/nekomata-7b-instruction: Full fine-tune

On the Hub · Downloads on Model Pulse