Big Transformer for Back-Translation

Open weights Facebook AI Research,Google Brain August 2018

No estimate

No hardware requirements for this model

This model's weights are open, but no parameter count has been published for it. Every memory and speed figure starts from that number, so we would rather show nothing than a fabricated estimate.

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
Facebook AI Research,Google Brain
Organisation type
Industry,Industry
Country
United States of America, France
Published
28 August 2018
Authors
Sergey Edunov, Myle Ott, Michael Auli, David Grangier

What it does

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

Domain
Language
Task
Translation
Approach
Supervised
Numerical format
FP16

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.

Training data
4,520,000,000 tokens

"Finally, for WMT English-German we train on all 226M available monolingual training sentences and perform 250K updates in 22.5 hours on 128 GPUs." We assume that 1 sentence have 15 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
4.8 × 10²⁰ FLOP

(128) * (1.25e14) * (27*3600 + 40*60) * (0.3) = 4.7808e20 (number of gpus) * (peak flops) * (seconds) * (assumed utilization rate) "We run experiments on DGX-1 machines with 8Nvidia V100 GPUs and machines are interconnected by Infiniband. Experiments are run on 16 machines and we perform 30K synchronous updates." "We also use the NCCL2 library [...] with 16-bit floating point operations" NCCL2 supported tensor core operations at 1.25e14 FLOP/s on a V100 for FP16 in section 5.6 we have "t…

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 Tesla V100 DGXS 16 GB
Chips used
128
Chip-hours
3,541
Wall-clock time
28 hours

"training updates in 27h 40min on 128 GPUs"

Power draw
66.2 kW
Compute cost
$2,442

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
Open — downloadable
Model access
Open weights (unrestricted)
Training code
Open source

Code and weights, MIT license: https://github.com/facebookresearch/fairseq/blob/main/examples/backtranslation/README.md

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,SOTA improvement

"Finally, we scale to hundreds of millions of monolingual sentences and achieve a new state of the art of 35 BLEU on the WMT'14 English-German test set. "

Record confidence
Likely
Citations
1,155

Sources

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

Reference
Understanding Back-Translation at Scale
Last updated
28 November 2025

What the numbers mean

Background

Big Transformer for Back-Translation was published by Facebook AI Research,Google Brain, in the country recorded as United States of America, during August 2018. The publishing organisation is categorised as industry,Industry.

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

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all.

What went into building it

Training it took a computation budget of roughly 4.8 × 10²⁰ FLOP, on hardware recorded as NVIDIA Tesla V100 DGXS 16 GB. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Training consumed a corpus of around 4,520,000,000 tokens of text.

The reason it appears in this catalogue at all: highly cited,SOTA improvement.

Answers

Big Transformer for Back-Translation — common questions

01

Big Transformer for Back-Translation— when was it released?

It was published in August 2018. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

02

Big Transformer for Back-Translation— what is it used for?

It works in the domain of Language, and is recorded as handling the task of translation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

03

Big Transformer for Back-Translation— where can I download it?

The weights are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

04

Big Transformer for Back-Translation— how much compute was used to train it?

Training consumed around 4.8 × 10²⁰ FLOP, on hardware recorded as NVIDIA Tesla V100 DGXS 16 GB. 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.

05

Big Transformer for Back-Translation— what GPU do I need to run it?

We cannot say. It has open weights, but no parameter count has been published for it, and every memory and speed calculation starts from that number. We would rather show nothing than a fabricated estimate.

06

Big Transformer for Back-Translation— is it open source?

Its weights are published, so it can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

07

Big Transformer for Back-Translation— how many parameters does it have?

No parameter count has been published for it, which is why no memory or speed figure appears on this page.

08

Big Transformer for Back-Translation— who created it?

It was published by Facebook AI Research,Google Brain, based in United States of America, an organisation categorised as industry,Industry.

Source

Original publication

Record last updated 28 November 2025

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.