How to Create Custom FEBRL Models for Your New Dataset in Docker Setup

puneet · 2024-07-22T07:53:28.668Z

hi team in the docker set up i want to run the febrl like example but on my new data set but im still figuring it out how to create custom own models for our new data set

Zingg Community

Sonal G.
·
Sania G.
Sania G.
·
https://docs.zingg.ai/zingg/stepbystep/configuration The above link has the details you need - you can make changes in the config.json file according to your requirements and play around with fields such as modelId etc. as per your data set Another option is to use the Python API - https://docs.zingg.ai/zingg/working-with-python/index#zingg-entity-resolution-python-package
puneet
·
Sania G. thanks for your reply actually i have a client demo on Zingg So what I understood so far is that i need to prepare the config.json file for my client data csv and run 1. findTrainingData 2. label and then we will get the new model in our zingg models folder ? and then that model id do we need to add in the config.json ?
puneet
·
How to create a new model inside zingg models directory for our new data set ?
puneet
·
Sania G. please guide
Sania G.
·
before running, change the modelId and then train and label data you can create a directory anywhere for the new model - it doesn't necessary need to be in the zingg models directory
puneet
·
Yes, Sania G. "Before running, change the model ID and then train and label the data" But how do I create a new model for my new data? All pretrained models are from 100 to 106 are for example data sets like febrl...etc i want to create new model for my new client data
puneet
·
i dont think so we can use modelId :100 for our new data set these are for example data sets ..Right ?
Sania G.
·
assign it a new value - apart from the used ones
puneet
·
Sania G. yes we will assign the new values in config.json but how to create a new model . so that we can pass this model id in the config.json
puneet
·
If schema and data is different then existing models will not work, we have to train our own model Sania G. is it right ?
👍1
puneet
·
Sania G. please reply im having a call with the client in next 2 hours
Sania G.
·
yes you will have to train you own model
puneet
·
can you please provide 2 to 3 liner steps how to create new model for my new data set ?
puneet
·
Sania G.

32 comments