Primer (GPT-3 XL-like 1.9B)
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
- Google Brain
- Organisation type
- Industry
- Country
- United States of America
- Published
- 24 January 2022
- Authors
- DavidR.So, WojciechMan ́ke, HanxiaoLiu, ZihangDai, NoamShazeer, QuocV.Le
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling/generation
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
- 1.9B
- Training data
- 2,000,000,000,000 tokens
- Batch size
- 2,000,000
"For instance, in a 1.9B parameter configuration similar to GPT-3 XL, Primer uses 1/3 of the training compute to achieve the same one-shot performance as Transformer"
"Each model is trained using batches of ∼2M tokens using 512 TPUv4 chips for ∼140 hours (∼71.8K total accelerator hours or ∼1M train steps). " 2 * 10^6 * 10^6 = 2*10^12 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
- 2.2 × 10²² FLOP
- How it was established
- Hardware,Operation counting
"Each model is trained using batches of ∼2M tokens using 512 TPUv4 chips for ∼140 hours (∼71.8K total accelerator hours or ∼1M train steps)." 71800 h * 0.3 [assumed utilization] * 275e12 FLOP/s * 3600s/h = 2.13246e+22 FLOP 6 FLOP / parameter / token * 1.9 * 10^9 parameters * 2 * 10^12 tokens = 2.28e+22 FLOP sqrt(2.13246e+22*2.28e+22) = 2.2049963e+22 FLOP
The training run
What it physically took to train: which chips, how many, for how long, and what that drew from the wall.
- Training hardware
- Google TPU v4
- Chips used
- 512
- Wall-clock time
- 140 hours
- Power draw
- 350.4 kW
"Each model is trained using batches of ∼2M tokens using 512 TPUv4 chips for ∼140 hours (∼71.8K total accelerator hours or ∼1M train steps)."
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
- Open source
Apache 2.0 We open source our models and several comparisons in T5 to help with reproducibility.: https://github.com/google-research/google-research/tree/master/primer seems like no model weights here
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Confident
- Citations
- 209
Sources
Where this record came from and when it was last checked.
- Reference
- Primer: Searching for Efficient Transformers for Language Modeling
- Last updated
- 25 May 2026
What the numbers mean
Where it came from
Primer (GPT-3 XL-like 1.9B) was published by Google Brain, in United States of America, in January 2022. industry is the category the publisher falls under.
It works in Language, and is recorded as doing language modeling/generation.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
What went into building it
Producing it required around 2.2 × 10²² FLOP of arithmetic, on Google TPU v4, which is a statement about the training budget rather than about inference.
The training set ran to roughly 2,000,000,000,000 tokens.
Answers
Primer (GPT-3 XL-like 1.9B) — common questions
How much compute was used to train Primer (GPT-3 XL-like 1.9B)?
Around 2.2 × 10²² FLOP, on Google TPU v4. 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 Primer (GPT-3 XL-like 1.9B)?
None. Primer (GPT-3 XL-like 1.9B) 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 Primer (GPT-3 XL-like 1.9B) open source?
No. Primer (GPT-3 XL-like 1.9B) has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does Primer (GPT-3 XL-like 1.9B) have?
Primer (GPT-3 XL-like 1.9B) has 1.9B parameters. "For instance, in a 1.9B parameter configuration similar to GPT-3 XL, Primer uses 1/3 of the training compute to achieve the same one-shot performance as Transformer". 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 Primer (GPT-3 XL-like 1.9B)?
Primer (GPT-3 XL-like 1.9B) was published by Google Brain, based in United States of America, categorised as industry.
When was Primer (GPT-3 XL-like 1.9B) released?
Primer (GPT-3 XL-like 1.9B) was published in January 2022. 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 Primer (GPT-3 XL-like 1.9B) used for?
Primer (GPT-3 XL-like 1.9B) works in Language, and is recorded as handling language modeling/generation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
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.