Four early open-source LLMs briefly ruled Chatbot Arena

Four early open-source models—Vicuna-13B, Guanaco-33B, Vicuna-33B, and WizardLM-70B—briefly dominated the Chatbot Arena, outperforming early commercial offerings. Vicuna-13B, trained for $300, pioneered the use of ChatGPT conversation data for fine-tuning and indirectly led to the creation of the Chatbot Arena platform. Guanaco-33B demonstrated the power of QLoRA for efficient fine-tuning on consumer hardware, a technique that revolutionized open-source model development. WizardLM-70B, developed by Microsoft, introduced the Evol-Instruct method for generating complex training data, though its successor, WizardLM-2, was mysteriously removed from public access shortly after its release. AI

IMPACT These early open-source models pioneered efficient training and data generation techniques, paving the way for today's advanced LLMs.

RANK_REASON The cluster details the history and technical innovations of early open-source LLMs that achieved high rankings on the Chatbot Arena benchmark. [lever_c_demoted from research: ic=1 ai=1.0]

Read on dev.to — LLM tag →

• Chatbot Arena
• ChatGPT
• Evol-Instruct
• GPT-4
• Guanaco-33B
• LMSYS
• LLaMA
• Microsoft
• QLoRA
• Vicuna-13B
• Vicuna-33B
• WizardLM-2
• WizardLM-70B

• model release
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →