Sparse all-MLP

Closed weights Meta AI 9.4B parameters April 2022

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
Meta AI
Organisation type
Industry
Country
United States of America
Published
14 April 2022
Authors
Ping Yu, Mikel Artexte, Myle Ott, Sam Shleifer, Hongyu Gong, Ves Stoyanov, Xian Li

What it does

The problem areas the model was built for. A model can carry several of each.

Domain
Language
Task
Language modeling
Approach
Self-supervised learning

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
9.4B

Table 2: "In Section 4.4, we run our large model (9.41B parameters)"

Training data
100,000,000,000 tokens

100B tokens (Table 2) so 75B words.

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
5.3 × 10²⁰ FLOP

112 hours on 32 V100 GPUs assumed 0.33 util rate 112 hours *3600 seconds / hour *0.33 utilization *32 gpus *125000000000000 FLOPs=532224000000000000000

How it was established
Hardware

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
Wall-clock time
112 hours

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.

Why it is tracked
SOTA improvement

"Through extensive evaluations on language modeling, we show that sMLP outperforms stateof-the-art sparse Transformer-based MoE models in terms of generalization and 2× improvement in training efficiency." not an absolute SOTA "Table 6. Zero-shot priming evaluation: we provide head-to-head comparison of our sMLP model with FLOPs-matched state-of-the-art sparse Transformers on six representative NLP tasks evaluated in GPT-3 in-context learning(Brown et al., 2020b)."

Record confidence
Confident
Citations
16

Sources

Where this record came from and when it was last checked.

Reference
Efficient Language Modeling with Sparse all-MLP
Last updated
25 May 2026

What the numbers mean

Where it came from

Sparse all-MLP was published by Meta AI, in United States of America, in April 2022. The organisation is categorised as industry.

It works in Language, and is recorded as doing language modeling.

Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.

How it was trained

The training run consumed about 5.3 × 10²⁰ FLOP, on NVIDIA V100. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

Around 100,000,000,000 tokens went into training it.

It is tracked in the underlying dataset for one reason in particular: sOTA improvement.

Answers

Sparse all-MLP — common questions

01

Is Sparse all-MLP open source?

No. Sparse all-MLP has not had its weights published, so it exists only as a service controlled by its owner.

02

How many parameters does Sparse all-MLP have?

Sparse all-MLP has 9.4B parameters. Table 2: "In Section 4.4, we run our large model (9.41B 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.

03

Who created Sparse all-MLP?

Sparse all-MLP was published by Meta AI, based in United States of America, categorised as industry.

04

When was Sparse all-MLP released?

Sparse all-MLP was published in April 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.

05

What is Sparse all-MLP used for?

Sparse all-MLP works in Language, and is recorded as handling language modeling. These are the areas it was designed around; they describe intent rather than a hard boundary.

06

How much compute was used to train Sparse all-MLP?

Around 5.3 × 10²⁰ FLOP, on 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.

07

What GPU do I need to run Sparse all-MLP?

None. Sparse all-MLP 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.

Source

Original publication

Record last updated 25 May 2026

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.