Ecco is a python library for creating interactive visualizations for Natural Language Processing models.
It provides multiple interfaces to aid the explanation and intuition of Transformer-based language models.
Ecco runs inside Jupyter notebooks. It is built on top of pytorch and transformers.
Input and Output Sequences
View the inputs and outputs (generated text) of language models broken down into tokens.
View the inputs and outputs (generated text) of language models broken down into tokens.
Input Saliency
How much did each input token contribute to producing the output token?
Neuron factors
Extracting underlying behaviour of neurons in a small number of factors with dimensionality reduction.
- Free software: BSD 3-Clause License
pip install ecco
To use the project:
import ecco
lm = ecco.from_pretrained('distilgpt2')
text= "The countries of the European Union are:\n1. Austria\n2. Belgium\n3. Bulgaria\n4."
output = lm.generate(text, generate=20, do_sample=True)