Llama and GPT-3 input durations compared to human language acquisition

A user added a circle for the Llama model to a visualization, noting that input from GPT-3 would represent 20,000 years of a child's input, while Llama's input would equate to 1.5 million years. This comparison highlights significant differences in how humans acquire language compared to AI models. AI

RANK_REASON This is a user-generated comparison of model input lengths, not a primary release or significant industry event.

Read on Mastodon — fosstodon.org →

• Bowerman
• GPT-3
• llama
• Wasstadt

• other

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