Subformer (122M)
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
- National Institute of Advanced Industrial Science and Technology (AIST),University of Tokyo
- Organisation type
- Academia,Academia
- Country
- Japan
- Published
- 1 January 2021
- Authors
- Machel Reid, Edison Marrese-Taylor, Yutaka Matsuo
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling
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
- 122M
- Training data
- tokens
- Epochs
- 70.29
- Batch size
- 24,576
122M (Table 5)
"we use 8 GPUs with 3072 tokens per GPU and an update frequency of 3" "warming up the learning rate to 1.0 for 16K iterations, and then annealing for 270K iterations" 8*3072*286000/100000000 = 70.29 epochs
8*3072
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
- 5.1 × 10¹⁸ FLOP
- How it was established
- Operation counting
6 FLOP/token/parameter * 122000000 parameters * 8 GPUs * 3072 tokens per batch per GPU * 286000 steps = 5.1450348e+18 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
- NVIDIA V100
- Chips used
- 8
- Power draw
- 4.9 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
code, but not for language modeling: https://github.com/machelreid/subformer/blob/master/README.md
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
- 60
- Benchmark data
- Subformer (122M)
Sources
Where this record came from and when it was last checked.
- Reference
- Subformer: Exploring Weight Sharing for Parameter Efficiency in Generative Transformers
- Last updated
- 25 May 2026
What the numbers mean
Background
Subformer (122M) was published by National Institute of Advanced Industrial Science and Technology (AIST),University of Tokyo, in the country recorded as Japan, during January 2021. The category the publisher falls under is academia,Academia.
It works in the domain of Language, and is recorded as performing the task of language modeling.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
How it was trained
The training run consumed about 5.1 × 10¹⁸ FLOP, on hardware recorded as NVIDIA V100. That figure measures what producing the model cost, and has no bearing on how fast it answers.
Answers
Subformer (122M) — common questions
Subformer (122M)— how much compute was used to train it?
Training consumed around 5.1 × 10¹⁸ FLOP, on hardware recorded as NVIDIA V100. 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.
Subformer (122M)— 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.
Subformer (122M)— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
Subformer (122M)— how many parameters does it have?
It has a parameter count of 122M. 122M (Table 5). 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.
Subformer (122M)— who created it?
It was published by National Institute of Advanced Industrial Science and Technology (AIST),University of Tokyo, based in Japan, an organisation categorised as academia,Academia.
Subformer (122M)— when was it released?
It was published in January 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.
Subformer (122M)— what is it used for?
It works in the domain of Language, and is recorded as handling the task of language modeling. 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.