AI models
KnGPT2
Training compute
7.9×10²⁰ FLOP
Parameters
83M
Published
Oct 15, 2021
KnGPT2 is an AI model developed by Huawei Noah's Ark Lab and McGill University (China and Canada), first published in October 2021. It works in the language domain, on tasks such as language modeling/generation.
Training it took an estimated 7.9×10²⁰ FLOP of compute (estimation method: operation counting). The model has 83,000,000 parameters. It was trained on roughly 853.3M datapoints.
Access: Unreleased. Its weights are not openly released. It is built on top of GPT-2 (124M). The reference paper has 42 citations. Epoch AI rates the confidence of this record as speculative.
Full record
- Organization
- Huawei Noah's Ark Lab, McGill University
- Country of organization
- China, Canada
- Domain
- Language
- Task
- Language modeling/generation
- Training compute
- 7.9×10²⁰ FLOP
- Compute estimation method
- Operation counting
- Parameters
- 83,000,000
- Dataset size
- 853.3M
- Training time
- 7 h
- Model accessibility
- Unreleased
- Open weights
- No
- Base model
- GPT-2 (124M)
- Citations
- 42
- Epoch confidence
- Speculative
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