CTR-BERT
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
- Amazon
- Organisation type
- Industry
- Country
- United States of America
- Published
- 6 December 2021
- Authors
- Aashiq Muhamed, Iman Keivanloo, Sujan Perera, James Mracek, Yi Xu, Qingjun Cui, Santosh Rajagopalan, Belinda Zeng, Trishul Chilimb
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Recommendation
- Task
- Click-through rate prediction
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
- 70M
- Training data
- 1,200,000,000 tokens
'CTR-BERTuses about 70 million parameters which can be trained on 8 A100 GPUs in less than a day (<1000 USD). '
more than 200M citations: 'Our CTR dataset is sampled from online traffic and is different from existing public CTR datasets in that it comprises of text features in addition to numeric/categorical features. We sample random train-test splits from 2020 online traffic. As The train-test splits are sampled from 2021 online traffic and balanced the same way as OOD data. The train set comprises 200 million data points and the test and validation sets comprise 25 million points each'
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
- 6.5 × 10¹⁹ FLOP
- How it was established
- Hardware
flops = (8) * (312 * 10**12) * (24 * 3600) * (0.3) (num gpu) * (peak flops) * (time in seconds) * (assumed utilization rate) 'CTR-BERT uses about 70 million parameters which can be trained on 8 A100 GPUs in less than a day (<1000 USD). '
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
- NVIDIA A100
- Chips used
- 8
- Power draw
- 6.4 kW
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
- Likely
- Citations
- 52
Sources
Where this record came from and when it was last checked.
- Reference
- CTR-BERT: Cost-effective knowledge distillation for billion-parameter teacher models
- Last updated
- 1 December 2025
What the numbers mean
About this model
CTR-BERT was published by Amazon, in the country recorded as United States of America, during December 2021. It comes out of an organisation categorised as industry.
It works in the domain of Recommendation, and is recorded as performing the task of click-through rate prediction.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
Training and provenance
The training run consumed about 6.5 × 10¹⁹ FLOP, on hardware recorded as NVIDIA A100. That figure measures what producing the model cost, and has no bearing on how fast it answers.
The training set ran to roughly 1,200,000,000 tokens of text.
Answers
CTR-BERT — common questions
CTR-BERT— how much compute was used to train it?
Training consumed around 6.5 × 10¹⁹ FLOP, on hardware recorded as NVIDIA A100. 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.
CTR-BERT— what GPU do I need to run it?
None. This 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.
CTR-BERT— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
CTR-BERT— how many parameters does it have?
It has a parameter count of 70M. 'CTR-BERTuses about 70 million parameters which can be trained on 8 A100 GPUs in less than a day (<1000 USD). '. 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.
CTR-BERT— who created it?
It was published by Amazon, based in United States of America, an organisation categorised as industry.
CTR-BERT— when was it released?
It was published in December 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.
CTR-BERT— what is it used for?
It works in the domain of Recommendation, and is recorded as handling the task of click-through rate prediction. 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.