Sandwich Transformer

Closed weights Allen Institute for AI,Facebook AI Research 209M parameters November 2019

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
Allen Institute for AI,Facebook AI Research
Organisation type
Research collective,Industry
Country
United States of America, France
Published
10 November 2019
Authors
Ofir Press, Noah A. Smith, Omer Levy

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
209M

209M "All of our experiments use the same hyperparameters as Baevski and Auli’s original model."

Training data
700,000,000 tokens

"while retaining the other architectural aspects and hyperparameter settings from Baevski and Auli (2019)" Baevski and Auli (2019): 286k steps in batches of 65,536 tokens. 286000*65536/700000000 = 27 epochs

Epochs
27
Batch size
65,536

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
2.4 × 10¹⁹ FLOP

6 FLOP / token / parameter * 209000000 parameters * 286000 steps * 65536 tokens per batch [same as Baevski and Auli (2019) = 2.3504093e+19 FLOP __________ in the Algorithmic progress paper they assumed 180 epochs (same as Baevski and Auli 2019 Transformer, but that one was trained on WT 103 not the book corpus) -> training compute was estimated to be 1.58E+20 FLOP

How it was established
Operation counting

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
Open (non-commercial)

non-commercial training and inference code: https://github.com/ofirpress/sandwich_transformer

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

"Sandwich transformers achieve state-of-the-art results on the enwik8 character-level language modeling dataset and on an additional word-level corpus, but have no significant effect on machine translation"

Record confidence
Speculative
Citations
97
Benchmark data
Sandwich Transformer

Sources

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

Reference
Improving Transformer Models by Reordering their Sublayers
Last updated
25 May 2026

What the numbers mean

About this model

Sandwich Transformer was published by Allen Institute for AI,Facebook AI Research, in the country recorded as United States of America, during November 2019. The category the publisher falls under is research collective,Industry.

It works in the domain of Language, and is recorded as performing the task of language modeling.

Its weights were never published, so it can only be reached through its provider. No graphics card changes that.

How it was trained

The training run consumed about 2.4 × 10¹⁹ FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.

The training set ran to roughly 700,000,000 tokens of text.

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

Answers

Sandwich Transformer — common questions

01

Sandwich Transformer— 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.

02

Sandwich Transformer— is it open source?

No. Its weights have not been published, so it exists only as a service controlled by its owner.

03

Sandwich Transformer— how many parameters does it have?

It has a parameter count of 209M. 209M "All of our experiments use the same hyperparameters as Baevski and Auli’s original model.". 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.

04

Sandwich Transformer— who created it?

It was published by Allen Institute for AI,Facebook AI Research, based in United States of America, an organisation categorised as research collective,Industry.

05

Sandwich Transformer— when was it released?

It was published in November 2019. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

06

Sandwich Transformer— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling. 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.

07

Sandwich Transformer— how much compute was used to train it?

Training consumed around 2.4 × 10¹⁹ FLOP. 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.

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.