perplexity-lenses / README.md
edugp's picture
Sync with data tooling repo, using edugp/kenlm models, updating viz to use quantiles for coloring and ad-hoc viz for the registry dataset
3c30fa3
|
raw
history blame
1.04 kB
metadata
title: Perplexity Lenses
emoji: 🌸
colorFrom: pink
colorTo: blue
sdk: streamlit
app_file: app.py
pinned: false

Installation:

Requires Python >= 3.7 and < 3.10

pip install .

Or with poetry

poetry install

Web App:

The app is hosted here. To run it locally:

python -m streamlit run app.py

CLI:

The CLI with no arguments defaults to running mc4 in Spanish. For full usage:

python cli.py --help

Example: Running on 1000 sentences extracted from Spanish OSCAR docs specifying all arguments:

python cli.py \
    --dataset oscar \
    --dataset-config unshuffled_deduplicated_es \
    --dataset-split train \
    --text-column text \
    --language es \
    --doc-type sentence \
    --sample 1000 \
    --dimensionality-reduction umap \
    --model-name distiluse-base-multilingual-cased-v1 \
    --output-file perplexity.html

Tests:

python -m unittest discover -s ./tests/ -p "test_*.py"