AI models
MedBERT
Training compute
9.5×10¹⁸ FLOP
Parameters
17M
Published
May 20, 2021
MedBERT is an AI model developed by Peng Cheng Laboratory and University of Texas at Houston (China and United States), first published in May 2021. It works in the medicine domain, on tasks such as medical diagnosis, text classification, prediction of hospital stay duration and 2 more.
Training it took an estimated 9.5×10¹⁸ FLOP of compute (estimation method: hardware). The model has 17,000,000 parameters. It was trained on roughly 14.6B datapoints. Training ran on 1 NVIDIA Tesla V100 DGXS 32 GB for about 168 hours. The compute alone is estimated at $62 in 2023 dollars.
Access: Unreleased. Its weights are not openly released. The reference paper has 835 citations. Epoch AI rates the confidence of this record as likely.
Full record
- Organization
- Peng Cheng Laboratory, University of Texas at Houston
- Country of organization
- China, United States
- Domain
- Medicine
- Task
- Medical diagnosis, Text classification, Prediction of hospital stay duration, Prediction of diabetic heart failure (DHF), Prediction of onset of pancreatic cancer (PaCa)
- Training compute
- 9.5×10¹⁸ FLOP
- Compute estimation method
- Hardware
- Parameters
- 17,000,000
- Dataset size
- 14.6B
- Training hardware
- NVIDIA Tesla V100 DGXS 32 GB
- Chips used
- 1
- Training time
- 168 h
- Chip-hours
- 168
- Training power draw
- 278 W
- Training cost (2023 USD)
- $62
- Numerical format
- FP32
- Model accessibility
- Unreleased
- Open weights
- No
- Citations
- 835
- Epoch confidence
- Likely
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