ReLU (NORB)
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
- University of Toronto
- Organisation type
- Academia
- Country
- Canada
- Published
- 15 June 2010
- Authors
- Nair, V., Hinton, G. E.
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Vision
- Task
- Object recognition
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
- 16.2M
- Training data
- 291,600 tokens
"The stereo-pair images are subsampled from their original resolution of 108 × 108 × 2 to 32 × 32 × 2 to speed up experiments [...] the architecture with the best results have 4000 units in the first layer and 2000 in the second [...] there are 58,320 test cases (9,720 cases per class) " So the architecture has (32*32*2+1)x4000 + (4000+1)*2000 + (2000+1)*58,320/9,720 parameters
"There are 291,600 training cases (48,600 cases per class) and 58,320 test cases (9,720 cases per class)."
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Why it is tracked
- Highly cited
- Citations
- 18,270
Sources
Where this record came from and when it was last checked.
- Reference
- Rectified linear units improve restricted boltzmann machines
- Last updated
- 11 February 2026
What the numbers mean
Background
ReLU (NORB) was published by University of Toronto, in Canada, in June 2010. It comes out of academia.
It works in Vision, and is recorded as doing object recognition.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
Training and provenance
It was trained on about 291,600 tokens of text.
Its inclusion criterion is highly cited.
Answers
ReLU (NORB) — common questions
Who created ReLU (NORB)?
ReLU (NORB) was published by University of Toronto, based in Canada, categorised as academia.
When was ReLU (NORB) released?
ReLU (NORB) was published in June 2010. 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 ReLU (NORB) used for?
ReLU (NORB) works in Vision, and is recorded as handling object recognition. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.
What GPU do I need to run ReLU (NORB)?
None. ReLU (NORB) 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 ReLU (NORB) open source?
The licensing for ReLU (NORB) was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.
How many parameters does ReLU (NORB) have?
ReLU (NORB) has 16.2M parameters. "The stereo-pair images are subsampled from their original resolution of 108 × 108 × 2 to 32 × 32 × 2 to speed up experiments [...] the architecture with the best results have 4000 units in the first layer and 2000 in the second [...] there are 58,320 test cases (9,720 cases per class) " So the architecture has (32*32*2+1)x4000 + (4000+1)*2000 + (2000+1)*58,320/9,720 parameters. 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.
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.