ISS

Closed weights Duke University,Microsoft 11.1M parameters September 2017

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
Duke University,Microsoft
Organisation type
Academia,Industry
Country
United States of America
Published
15 September 2017
Authors
Wei Wen, Yuxiong He, Samyam Rajbhandari, Minjia Zhang, Wenhan Wang, Fang Liu, Bin Hu, Yiran Chen, Hai Li

What it does

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

Domain
Language
Task
Language modeling
Numerical format
FP32

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
11.1M

11.1M (Table 2)

Training data
929,000 tokens

"All models are trained from scratch for 55 epochs."

Epochs
55

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
3.4 × 10¹⁵ FLOP

6 FLOP / parameter / token * 11100000 parameters * 929000 tokens * 55 epochs = 3.402927e+15 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 source

code (apache): https://github.com/wenwei202/iss-rnns/tree/master/ptb

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

"Moreover, ISS learning can find a smaller RHN model with width 726, meanwhile improve the state-of-the-art perplexity as shown by the second entry in Table 2."

Record confidence
Confident
Citations
146
Benchmark data
ISS

Sources

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

Reference
Learning Intrinsic Sparse Structures within Long Short-Term Memory
Last updated
28 November 2025

What the numbers mean

About this model

ISS was published by Duke University,Microsoft, in United States of America, in September 2017. It comes out of academia,Industry.

It works in Language, and is recorded as doing 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 3.4 × 10¹⁵ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

It was trained on about 929,000 tokens of text.

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

Answers

ISS — common questions

01

How many parameters does ISS have?

ISS has 11.1M parameters. 11.1M (Table 2). 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.

02

Who created ISS?

ISS was published by Duke University,Microsoft, based in United States of America, categorised as academia,Industry.

03

When was ISS released?

ISS was published in September 2017. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

04

What is ISS used for?

ISS works in Language, and is recorded as handling language modeling. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

05

How much compute was used to train ISS?

Around 3.4 × 10¹⁵ 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.

06

What GPU do I need to run ISS?

None. ISS 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.

07

Is ISS open source?

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

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.