AI models
PocketGen
Training compute
2.1×10¹⁹ FLOP
Parameters
7.9M
Published
Sep 23, 2024
PocketGen is an AI model developed by University of Science and Technology of China (USTC), Hefei Comprehensive National Science Center, Harvard University, Broad Institute and Harvard Data Science Initiative (China and United States), first published in September 2024. It works in the biology domain, on tasks such as protein-ligand contact prediction.
Training it took an estimated 2.1×10¹⁹ FLOP of compute (estimation method: hardware). The model has 7,900,000 parameters. Training ran on 1 NVIDIA A100 for about 48 hours.
The reference paper has 15 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- University of Science and Technology of China (USTC), Hefei Comprehensive National Science Center, Harvard University, Broad Institute, Harvard Data Science Initiative
- Country of organization
- China, United States
- Domain
- Biology
- Task
- Protein-ligand contact prediction
- Training compute
- 2.1×10¹⁹ FLOP
- Compute estimation method
- Hardware
- Parameters
- 7,900,000
- Training hardware
- NVIDIA A100
- Chips used
- 1
- Training time
- 48 h
- Training power draw
- 433 W
- Citations
- 15
- Epoch confidence
- Confident
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.
← All ai models