RQ-Transformer (3.8B params ImageNet dataset)
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
- Kakao,POSTECH
- Organisation type
- Industry,Academia
- Country
- Korea (Republic of)
- Published
- 3 March 2022
- Authors
- Doyup Lee, Chiheon Kim, Saehoon Kim, Minsu Cho, Wook-Shin Han
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Vision, Image generation
- Task
- Text-to-image
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
- 3.8B
- Training data
- 327,680,000 tokens
- Epochs
- 50
3822M from Table 5 - ImageNet rows
size of ImageNet
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
- 2.9 × 10²⁰ FLOP
- How it was established
- Hardware
"We use eight NVIDIA A100 GPUs to train RQ-Transformer of 1.4B parameters, and four GPUs to train RQ-Transformers of other sizes. The training time is <9 days for LSUN-cat, LSUN-bedroom, <4.5 days for ImageNet, and CC-3M, and <1 day for LSUN-church and FFHQ. " Taken literally, suggests 3.8B model used fewer GPUs than 1.4B. This seems likely to be a typo, so I assume details are meant to be given for largest model (3.8B). (8) * (3.12e14) * (4.5 * 24 * 3600) * (0.3) = 2.911e20 (num gpu) * (peak …
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
- 4
- Chip-hours
- 432
- Wall-clock time
- 108 hours
- Power draw
- 3.2 kW
States 4.5 days for ImageNet for 1.4B model, but this is probably a typo. I expect these details were given for the largest model. "We use eight NVIDIA A100 GPUs to train RQ-Transformer of 1.4B parameters, and four GPUs to trainRQ-Transformers of other sizes. The training time is <9 days for LSUN-cat, LSUN-bedroom, <4.5 days for ImageNet, and CC-3M, and <1 day for LSUN-church and FFHQ. "
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
- 763
Sources
Where this record came from and when it was last checked.
- Reference
- Autoregressive Image Generation using Residual Quantization
- Last updated
- 25 May 2026
What the numbers mean
What this model is
RQ-Transformer (3.8B params ImageNet dataset) was published by Kakao,POSTECH, in the country recorded as Korea (Republic of), during March 2022. The category the publisher falls under is industry,Academia.
It works in the domain of Vision, Image generation, and is recorded as performing the task of text-to-image.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
Training and provenance
Training it took a computation budget of roughly 2.9 × 10²⁰ 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 327,680,000 tokens of text.
Answers
RQ-Transformer (3.8B params ImageNet dataset) — common questions
RQ-Transformer (3.8B params ImageNet dataset)— when was it released?
It was published in March 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.
RQ-Transformer (3.8B params ImageNet dataset)— what is it used for?
It works in the domain of Vision, Image generation, and is recorded as handling the task of text-to-image. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
RQ-Transformer (3.8B params ImageNet dataset)— how much compute was used to train it?
Training consumed around 2.9 × 10²⁰ 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.
RQ-Transformer (3.8B params ImageNet dataset)— 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.
RQ-Transformer (3.8B params ImageNet dataset)— is it open source?
The licensing was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.
RQ-Transformer (3.8B params ImageNet dataset)— how many parameters does it have?
It has a parameter count of 3.8B. 3822M from Table 5 - ImageNet rows. 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.
RQ-Transformer (3.8B params ImageNet dataset)— who created it?
It was published by Kakao,POSTECH, based in Korea (Republic of), an organisation categorised as industry,Academia.
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.