Guanaco vs. WizardLM

Overview

Guanaco is an LLM based off the QLoRA 4-bit finetuning method developed by Tim Dettmers et. al. in the UW NLP group. Guanaco achieves 99% ChatGPT performance on the Vicuna benchmark.

Guanaco is an LLM that uses a finetuning method called LoRA that was developed by Tim Dettmers et. al. in the UW NLP group. With QLoRA, it becomes possible to finetune up to a 65B parameter model on a 48GB GPU without loss of performance relative to a 16-bit model. The Guanaco model family outperforms all previously released models on the Vicuna benchmark. However, given the models are based off of the LLaMA model family, commercial use is not permitted.

Initial release: 2023-05-23

Reference

https://github.com/artidoro/qlora

Further Reading

• https://arxiv.org/abs/2305.14314
• https://twitter.com/Tim_Dettmers/status/1661379354507476994

Overview

WizardLM is a variant of LLaMA trained with complex instructions.

WizardLM is a LLM based on LLaMA trained using a new method, called Evol-Instruct, on complex instruction data. By using AI to "evolve" instructions, WizardLM outperforms similar LLaMA-based LLMs trained on simpler instruction data. On the 6th of July, 2023, WizardLM V1.1 was released with significantly improved performance, and as of 15 April 2024, WizardLM-2 was released with state-of-the-art performance.

Initial release: 2023-05-26

Reference

https://github.com/nlpxucan/WizardLM

Further Reading

• https://huggingface.co/WizardLM
• https://arxiv.org/abs/2304.12244
• https://www.reddit.com/r/LocalLLaMA/comments/142iw20/official_wizardlm30b_v10_released_can_beat/