KnGPT2
No estimate
No hardware requirements for this model
The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- Huawei Noah's Ark Lab,McGill University
- Organisation type
- Industry,Academia
- Country
- China, Canada
- Published
- 15 October 2021
- Authors
- Ali Edalati, Marzieh Tahaei, Ahmad Rashid, Vahid Partovi Nia, James J. Clark, Mehdi Rezagholizadeh
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling/generation
- Base model
- GPT-2 (124M)
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Parameters
- 83M
- Training data
- 853,333,333 tokens
- Epochs
- 1
"The KnGPT2 model is compressed from the GPT-2Small Radford et al. (2019) model. GPT-2Small is 124 million parameters. Our baseline is DistilGPT2 which has about 82 million parameters so our KnGPT2 model is compressed to the same size (83 million parameters) for a fair comparison"
"KnGPT2 is pre-trained on 10% of OpenWebText" 3.2 GB (Table 3) 3.2GB * 200*10^6 words per GB * 4/3 tokens per word = 853333333 tokens
Training compute
The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.
- Training compute
- 7.9 × 10²⁰ FLOP
- How it was established
- Operation counting
- Fine-tuning compute
- 4.2 × 10¹⁷ FLOP
7.936e+20 FLOP [base model, "Speculative" confidence"] + 4.2496e+17 FLOP [fine-tune] = 7.9402496e+20 FLOP
6 FLOP / parameter / token * 83000000 parameters * 853333333 tokens = 4.2496e+17 FLOP
The training run
What it physically took to train: which chips, how many, for how long, and what that drew from the wall.
- Wall-clock time
- 7 hours
6.5 hours (Table 3)
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Closed — provider access only
- Model access
- Unreleased
- Training code
- Unreleased
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Speculative
- Citations
- 42
- Benchmark data
- KnGPT2
Sources
Where this record came from and when it was last checked.
- Reference
- Kronecker Decomposition for GPT Compression
- Last updated
- 25 May 2026
What the numbers mean
Background
KnGPT2 was published by Huawei Noah's Ark Lab,McGill University, in China, in October 2021. The organisation is categorised as industry,Academia.
It works in Language, and is recorded as doing language modeling/generation.
Its starting point was GPT-2 (124M) — most models at this scale are adapted from an existing base rather than built from nothing.
Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.
What went into building it
The training run consumed about 7.9 × 10²⁰ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
Around 853,333,333 tokens went into training it.
Answers
KnGPT2 — common questions
How much compute was used to train KnGPT2?
Around 7.9 × 10²⁰ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.
What GPU do I need to run KnGPT2?
None. KnGPT2 is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.
Is KnGPT2 open source?
No. KnGPT2 has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does KnGPT2 have?
KnGPT2 has 83M parameters. "The KnGPT2 model is compressed from the GPT-2Small Radford et al. (2019) model. GPT-2Small is 124 million parameters. Our baseline is DistilGPT2 which has about 82 million parameters so our KnGPT2 model is compressed to the same size (83 million parameters) for a fair comparison". That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.
Who created KnGPT2?
KnGPT2 was published by Huawei Noah's Ark Lab,McGill University, based in China, categorised as industry,Academia.
When was KnGPT2 released?
KnGPT2 was published in October 2021. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.
What is KnGPT2 used for?
KnGPT2 works in Language, and is recorded as handling language modeling/generation. These are the areas it was designed around; they describe intent rather than a hard boundary.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.