liveBook · Manning

SlovenBERTcina

As explained in LLMs in Production

A language model trained on Slovak-language data for strong performance on selected language tasks.

A general-purpose model is not always the best choice for a narrow language task. SlovenBERTcina was trained on Slovak-language data and reportedly outperformed other available models on the tasks it was trained for, without setting computational records. section 1.2.2

Building the Slovak dataset

Training data came from the Slovak National Corpus, scraped with permission, and resources including OSCAR and Europarl. The result illustrates the appeal of building a focused model when task performance and control over data matter more than broad capability. section 1.2.2

The performance claim is specific to the tasks trained on; no benchmark results or training procedure are given. Use it as a reason to evaluate a focused model against your own tasks, not as a guarantee that custom training will win. section 1.2.2

How LLMs in Production covers it

How deep it goes
Names training-data sources and reports comparative performance, but supplies no training procedure or benchmark results.
Examples
Uses a Slovak-language dataset assembled from the Slovak National Corpus, OSCAR, and Europarl.
What you can do after
Evaluate/understand it as a case for comparing focused models on the tasks that matter to you.
How it connects
Connects BERT-era model customization with the Slovak National Corpus, OSCAR, Europarl, and data privacy.
Cover of LLMs in Production
liveBookmanning.com
Read LLMs in Production on liveBook Get the book at manning.com

This explainer was generated by Manning's knowledge graph from the text of LLMs in Production. The section links point to the parts of the book it draws on.