macadeliccc/SmaugDolphin-34B-slerp

Full fine-tune vs abacusai/Smaug-34B-v0.1. Fine-tune: 98% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 1/1, Norms 0/97, MLP 141/144, Attention 188/192, lm_head 1/1

2 derivatives checked

• mradermacher/SmaugDolphin-34B-slerp-i1-GGUF: Quantization
• mradermacher/SmaugDolphin-34B-slerp-GGUF: Quantization

On the Hub · Downloads on Model Pulse