Question

If and when this graph is extended to 10^14 parameter models trained on 10^14 elapsed tokens of similar-quality data, will the 10^14 parameter learning curve have slowed down substantially?

22 comments22

Total Forecasters29

Community Prediction29%

Make a Prediction

50%

community: 29%

Authors:

kokotajlod

Opened:

Closes:

Scheduled resolution:

AI Training and Compute

AI Technical Benchmarks

Computing and Math

Artificial Intelligence

Machine learning

Artificial intelligence

How many billions of parameters will the largest machine learning model trained before 2030 have?

2.43M

Will AI capabilities plateau with the next generation (GPT-5, etc.) of language models?

Will a dense machine learning model with at least 100 trillion parameters be trained before 2026?

28%

Comments

? comments

Authors:

kokotajlod

Opened:

Closes:

Scheduled resolution:

AI Training and Compute

AI Technical Benchmarks

Computing and Math

Artificial Intelligence

Machine learning

Artificial intelligence

The Long Road to Genuine AI Mastery

Time•Sep 12, 2024

Do Large Language Models Have a Subconscious?

Psychology Today•Sep 12, 2024

OpenAI releases o1, its first model with ‘reasoning’ abilities

Verge•Sep 13, 2024

Learn more about Metaculus NewsMatch

How many billions of parameters will the largest machine learning model trained before 2030 have?

2.43M

Will AI capabilities plateau with the next generation (GPT-5, etc.) of language models?

Will a dense machine learning model with at least 100 trillion parameters be trained before 2026?

28%