Comparing Models

 

Overview

In the Comparing Models notebook, we’re evaluating en.sentiment.imdb.glove model from JSL and lvwerra/distilbert-imdb from HF which are models trained for text classification task. The notebook showcases how we can get a view of how the models compare. The final report output is different than normal and includes both models’s results.

Open in Collab

Category Hub Task Open In Colab
Multiple Model Comparison Hugging Face/John Snow Labs/Spacy NER/Text-Classification Open In Colab

Config Used

tests:
  defaults:
    min_pass_rate: 1.0
  robustness:
    add_typo:
      min_pass_rate: 0.7
    american_to_british:
      min_pass_rate: 0.7
  accuracy:
    min_micro_f1_score:
      min_score: 0.7
  bias:
    replace_to_female_pronouns:
      min_pass_rate: 0.7
    replace_to_low_income_country:
      min_pass_rate: 0.7
  fairness:
    min_gender_f1_score:
      min_score: 0.6
  representation:
    min_label_representation_count:
      min_count: 50
Last updated