✨
AI Summary
- Wing Lian's Axolotl is emergent finetuning library of choice for open models like Llama 2 and Mistral 7B, enabling customization and ownership
- Finetuning enables privacy, control, and performance improvements (censorship, multiturn chat, chain of thought) while matching larger models with smaller architectures
- Sub-industry of finetuned variants growing around good open models to improve benchmark performance and reduce inference costs/latency