Markov-driven POS tagger

Closed weights EURECOM 2.4M parameters June 1994

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
EURECOM
Organisation type
Academia
Country
France
Published
1 June 1994
Authors
Bernard Merialdo

What it does

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

Domain
Language
Task
Part-of-speech tagging

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
2.4M

"The total number of free parameters is then: (Nw - 1).NT + (NT - 1).NT.NT." Where: Nw= Vocabulary size NT = Number of tags "In the treebank 159 different tags are used. These tags were projected on a smaller system of 76 tags designed by Evelyne Tzoukermann and Peter Brown (see Appendix). The results quoted in this paper all refer to this smaller system" So NT = 76 https://www.aclweb.org/anthology/J94-2001/ There is no direct reference to Nw, but the data is from "Lexicon and grammar in prob…

Training data
tokens

"We use the "treebank" data described in Beale (1988). It contains 42,186 sentences (about one million words) from the Associated Press." https://www.aclweb.org/anthology/J94-2001.pdf

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Citations
788

Sources

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

Reference
Tagging English Text with a Probabilistic Model
Last updated
28 November 2025

What the numbers mean

Background

Markov-driven POS tagger was published by EURECOM, in France, in June 1994. The organisation is categorised as academia.

It works in Language, and is recorded as doing part-of-speech tagging.

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

Answers

Markov-driven POS tagger — common questions

01

Who created Markov-driven POS tagger?

Markov-driven POS tagger was published by EURECOM, based in France, categorised as academia.

02

When was Markov-driven POS tagger released?

Markov-driven POS tagger was published in June 1994. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

03

What is Markov-driven POS tagger used for?

Markov-driven POS tagger works in Language, and is recorded as handling part-of-speech tagging. 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.

04

What GPU do I need to run Markov-driven POS tagger?

None. Markov-driven POS tagger 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.

05

Is Markov-driven POS tagger open source?

The licensing for Markov-driven POS tagger was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.

06

How many parameters does Markov-driven POS tagger have?

Markov-driven POS tagger has 2.4M parameters. "The total number of free parameters is then: (Nw - 1).NT + (NT - 1).NT.NT." Where: Nw= Vocabulary size NT = Number of tags "In the treebank 159 different tags are used. These tags were projected on a smaller system of 76 tags designed by Evelyne Tzoukermann and Peter Brown (see Appendix). The results quoted in this paper all refer to this smaller system" So NT = 76 https://www.aclweb.org/anthology/J94-2001/ There is no direct reference to Nw, but the data is from "Lexicon and grammar in probabilistic tagging of written English." which says "(the new CLAWS lexicón has almost 26,500 entries)" So tentatively Nw=26500 https://dl.acm.org/doi/10.3115/982023.982049. 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.

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.