AI models
Decaying Fast Weights Transformer (WT-103)
Training compute
7.9×10¹⁹ FLOP
Parameters
242M
Published
Oct 9, 2022
Decaying Fast Weights Transformer (WT-103) is an AI model developed by Jenni (Singapore), first published in October 2022. It works in the language domain, on tasks such as language modeling.
Training it took an estimated 7.9×10¹⁹ FLOP of compute (estimation method: hardware). The model has 242,000,000 parameters. It was trained on roughly 103M datapoints. Training ran on 1 NVIDIA A40 PCIe for about 35 hours.
Access: Unreleased. Its weights are not openly released. It is built on top of Transformer (Adaptive Input Embeddings) WT103. The reference paper has 30 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Jenni
- Country of organization
- Singapore
- Domain
- Language
- Task
- Language modeling
- Training compute
- 7.9×10¹⁹ FLOP
- Compute estimation method
- Hardware
- Parameters
- 242,000,000
- Dataset size
- 103M
- Training hardware
- NVIDIA A40 PCIe
- Chips used
- 1
- Training time
- 35 h
- Training power draw
- 330 W
- Model accessibility
- Unreleased
- Open weights
- No
- Base model
- Transformer (Adaptive Input Embeddings) WT103
- Citations
- 30
- 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