core.example
Main Example
This is an entrypoint file runnable via
./example.py
Run a sample setup
- consiting of fine tuning or training each model,
- evaluating the performance, and
- running a sample prediction.
This is a dummy sample, use core.experiments
to run real
experiments.
=> It may be also interesting to take a look at this source →
=> This code heavily uses core.model.exec.Executor
View Source
#!/usr/bin/env python3 """ # Main Example > This is an *entrypoint file* runnable via ``./example.py`` Run a sample setup - consiting of fine tuning or training each model, - evaluating the performance, and - running a sample prediction. **This is a dummy sample, use `core.experiments` to run *real* experiments.** => It may be also interesting to take a look at this source → => This code heavily uses `core.model.exec.Executor` """ # set seed from core.utils import Random, print_info Random.set_seed(123) # load corpora and models from core.corpus import Wiktionary, TwentyNews from core.model.transformer import IsSCDBert, IsNextSCDBert, SelectSCDBert, GivenSCDFindTextBert, GivenTextFindSCDBert from core.model.scdmatrix import iSCDMatrix, MPSCDMatrix # load executor from core.model import Executor if __name__ == "__main__": print_info() executions = [ (iSCDMatrix, ["The bison is cool!"]), (IsSCDBert, ["The bison is cool!"]), (IsNextSCDBert, ["The bison is cool!", "The bison is an animal."]), (SelectSCDBert, [["The bison is cool!"] * 4, ["A car has an engine.", "The computer calculates results.", "The bison is an animal.", "Somewhere is someone."]]), (GivenTextFindSCDBert, ["The bison is cool!", "A car has an engine. The computer calculates results. The bison is an animal. Somewhere is someone."]), (GivenSCDFindTextBert, ["The bison is an animal.", "I see cars. We use the computer to do research. The bison is cool!"]), (MPSCDMatrix, ["The bison is cool!", ["A car has an engine.", "The computer calculates results.", "The bison is an animal.", "Somewhere is someone."]]) ] for model, args in executions: corpus, _ = TwentyNews(subgroups=['misc-forsale']).split(percentages=[0.05, 0.95]) e = Executor( model, Wiktionary, corpus, percentage_train=0.8 ) print(e.exec(save_results=False)) print(e.predict(*args))