AI models
KN5 LM + RNN 400/10 (WSJ)
Training compute
1.7×10¹⁶ FLOP
Parameters
22.2M
Published
Sep 26, 2010
KN5 LM + RNN 400/10 (WSJ) is an AI model developed by Brno University of Technology and Johns Hopkins University (Czechia and United States), first published in September 2010. It works in the speech domain, on tasks such as transcription. It counts among the frontier models: the systems trained with the most compute of their moment.
Training it took an estimated 1.7×10¹⁶ FLOP of compute (estimation method: operation counting). The model has 22,160,000 parameters. It was trained on roughly 6.4M datapoints.
The reference paper has 6,038 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Brno University of Technology, Johns Hopkins University
- Country of organization
- Czechia, United States
- Domain
- Speech
- Task
- Transcription
- Training compute
- 1.7×10¹⁶ FLOP
- Compute estimation method
- Operation counting
- Parameters
- 22,160,000
- Dataset size
- 6.4M
- Training time
- 504 h
- Citations
- 6,038
- Epoch confidence
- Confident
More from Brno University of Technology,Johns Hopkins University
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