Appearance
Models
A model is what a processing job runs with. Every job that recognises, detects or extracts takes a model ID. This section explains the model families, which jobs they serve, and how to find model IDs. Models you train yourself in Transkribus are addressed exactly like public ones.
Model IDs come from the public models list on the Transkribus website or from your own models in the Transkribus app. Two to start with: Text Titan II, the current super model, is 579509; German_Kurrent_XIX-XX_M6-2, a PyLaia model for German Kurrent, is 38230.
Look up a model
Returns a model's name, provider, language and error rates. No token is needed for public models.
GET /models/{type}/{id}
GET
https://transkribus.eu/TrpServer/rest/models/{type}/{id}List the models you can use
type is the model family, text or layout; public models and the ones shared with your collections are included.
GET /models/{type}
GET
https://transkribus.eu/TrpServer/rest/models/{type}Parameters
Path Parameters
type*
Type
Requiredstring
Query Parameters
collId
Type
integer
Format
"int32"userid
Type
integer
Format
"int32"filter
Type
string
releaseLevel
Type
array
dataSetReleaseLevel
Type
array
prov
Type
string
centuries
Type
array
isoLanguages
Type
array
scriptTypes
Type
array
creator
Type
string
publisher
Type
string
docType
Type
string
featured
Type
boolean
application
Type
string
slug
Type
string
labels
Type
array
datasetId
Type
integer
Format
"int32"datasetVersionId
Type
integer
Format
"int32"minNrOfTrainGtPages
Type
integer
Format
"int32"minNrOfValidationGtPages
Type
integer
Format
"int32"facetLimit
Type
integer
Maximum
500Minimum
10Format
"int32"Default
10minFacetCount
Type
integer
Maximum
10Minimum
1Format
"int32"Default
2index
Type
integer
Format
"int32"Default
0nValues
Type
integer
Format
"int32"Default
-1sortColumn
Type
string
sortDirection
Type
string
Responses
default response
Training
Training a model from ground truth is a job on the Transkribus REST API, one call per model family: PyLaia for text, layout, tables. The operations are in the Training group beneath this page in the sidebar; the material comes from a dataset.
- Model types: super models vs. PyLaia models, and when to use which
- Look up a model and list your models, below