Mesh-TensorFlow Transformer 4.9B (language)
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
- Google Brain
- Organisation type
- Industry
- Country
- United States of America
- Published
- 5 November 2018
- Authors
- Noam Shazeer, Youlong Cheng, Niki Parmar, Dustin Tran, Ashish Vaswani, Penporn Koanantakool, Peter Hawkins, Hyoukjoong Mingsheng Lee, Cliff Hong, Ryan Young, Blake Sepassi, Hechtman
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling/generation, Translation
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
- 4.9B
- Training data
- 5,000,000,000 tokens
- Epochs
- 10
4.9B from section 9.1 : ''The largest model (4.9B parameters) took 13 hours to train on a 512-core TPUv2 cluster.'
from section 9.1. Experiments done on a "billion word benchmark" and a 5B token wikipedia dataset. At 4/3 tokens per word, 1.3B tokens in the first.
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
- 1.6 × 10²⁰ FLOP
- How it was established
- Hardware
flops = (256) * ( 45 * 10**12) * (13 * 3600) * (0.3) = 1.6e20 (num gpu) * (peak flops) * (time in seconds) * (assumed utilization rate) from section 9.1 : ''The largest model (4.9B parameters) took 13 hours to train on a 512-core TPUv2 cluster.' from https://en.wikipedia.org/wiki/Tensor_Processing_Unit 45TFLOPs per chips
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
- Google TPU v2
- Chips used
- 256
- Chip-hours
- 3,328
- Wall-clock time
- 13 hours
- Power draw
- 148.1 kW
- Compute cost
- $935
from section 9.1 "For the billion-word language modeling benchmark, we trained the models for 10 epochs. The largest model (4.9B parameters) took 13 hours to train on a 512-core TPUv2 cluster."
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 source
code here, apache license: https://github.com/tensorflow/mesh/tree/master/mesh_tensorflow/transformer https://github.com/tensorflow/mesh/blob/master/mesh_tensorflow/transformer/transformer.py
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
- Record confidence
- Confident
- Citations
- 437
'Using TPU meshes of up to 512 cores, we train Transformer models with up to 5 billion parameters, surpassing state of the art results on WMT'14 English-to-French translation task and the one-billion-word language modeling benchmark.'
Sources
Where this record came from and when it was last checked.
- Reference
- Mesh-TensorFlow: Deep Learning for Supercomputers
- Last updated
- 25 May 2026
What the numbers mean
About this model
Mesh-TensorFlow Transformer 4.9B (language) was published by Google Brain, in the country recorded as United States of America, during November 2018. The publishing organisation is categorised as industry.
It works in the domain of Language, and is recorded as performing the task of language modeling/generation, Translation.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
How it was trained
Training it took a computation budget of roughly 1.6 × 10²⁰ FLOP, on hardware recorded as Google TPU v2. That figure measures what producing the model cost, and has no bearing on how fast it answers.
The training set ran to roughly 5,000,000,000 tokens of text.
The reason it appears in this catalogue at all: sOTA improvement.
Answers
Mesh-TensorFlow Transformer 4.9B (language) — common questions
Mesh-TensorFlow Transformer 4.9B (language)— 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.
Mesh-TensorFlow Transformer 4.9B (language)— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
Mesh-TensorFlow Transformer 4.9B (language)— how many parameters does it have?
It has a parameter count of 4.9B. 4.9B from section 9.1 : ''The largest model (4.9B parameters) took 13 hours to train on a 512-core TPUv2 cluster.'. 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.
Mesh-TensorFlow Transformer 4.9B (language)— who created it?
It was published by Google Brain, based in United States of America, an organisation categorised as industry.
Mesh-TensorFlow Transformer 4.9B (language)— when was it released?
It was published in November 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.
Mesh-TensorFlow Transformer 4.9B (language)— what is it used for?
It works in the domain of Language, and is recorded as handling the task of language modeling/generation, Translation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
Mesh-TensorFlow Transformer 4.9B (language)— how much compute was used to train it?
Training consumed around 1.6 × 10²⁰ FLOP, on hardware recorded as Google TPU v2. 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.
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.