DITTO

Closed weights Tsinghua University,Apple,Westlake University,Chinese University of Hong Kong (CUHK) 750M parameters June 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
Tsinghua University,Apple,Westlake University,Chinese University of Hong Kong (CUHK)
Organisation type
Academia,Industry,Academia,Academia
Country
China, United States of America, Hong Kong
Published
6 June 2022
Authors
Jin Xu, Xiaojiang Liu, Jianhao Yan, Deng Cai, Huayang Li, Jian Li

What it does

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

Domain
Language
Task
Language modeling/generation, Text summarization
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.

Parameters
750M

"We train a Transformer model (750M parameters, similar to GPT-2 Large)" "Specifically, we use a 16-layer Transformer with 8 attention heads, hidden size 1024 and fully-connected dimension 4096."

Training data
103,000,000 tokens
Epochs
7.16

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
3.3 × 10¹⁸ FLOP

6 FLOP / token / parameter * 160000 steps * 3 samples per batch * 1536 tokens per sample * 750000000 parameters = 3.31776 × 10^18 FLOP

How it was established
Operation counting

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
Chips used
8
Power draw
4.8 kW

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

open code https://github.com/Jxu-Thu/DITTO training: https://github.com/Jxu-Thu/DITTO/blob/main/train.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

Achieves SOTA on CNN/DailyMail by fine-tuning and improving on BART-large, which is SOTA

Record confidence
Confident
Citations
109
Benchmark data
DITTO

Sources

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

Reference
Learning to Break the Loop: Analyzing and Mitigating Repetitions for Neural Text Generation
Last updated
25 May 2026

What the numbers mean

What this model is

DITTO was published by Tsinghua University,Apple,Westlake University,Chinese University of Hong Kong (CUHK), in China, in June 2022. The organisation is categorised as academia,Industry,Academia,Academia.

It works in Language, and is recorded as doing language modeling/generation, Text summarization.

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

Training and provenance

Training it took roughly 3.3 × 10¹⁸ FLOP of computation, on NVIDIA V100 — a measure of what producing the model cost, not of how fast it answers.

The training set ran to roughly 103,000,000 tokens.

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

Answers

DITTO — common questions

01

How much compute was used to train DITTO?

Around 3.3 × 10¹⁸ 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.

02

What GPU do I need to run DITTO?

None. DITTO 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.

03

Is DITTO open source?

No. DITTO has not had its weights published, so it exists only as a service controlled by its owner.

04

How many parameters does DITTO have?

DITTO has 750M parameters. "We train a Transformer model (750M parameters, similar to GPT-2 Large)" "Specifically, we use a 16-layer Transformer with 8 attention heads, hidden size 1024 and fully-connected dimension 4096.". 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.

05

Who created DITTO?

DITTO was published by Tsinghua University,Apple,Westlake University,Chinese University of Hong Kong (CUHK), based in China, categorised as academia,Industry,Academia,Academia.

06

When was DITTO released?

DITTO was published in June 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.

07

What is DITTO used for?

DITTO works in Language, and is recorded as handling language modeling/generation, Text summarization. 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.

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.