Awesome

few-shot-text-classification

Code for reproducing the results from the paper Few Shot Text Classification with a Human in the Loop. This repo contains the SIF code from the (Arora et al, 2017) paper "A Simple but Tough-to-Beat Baseline for Sentence Embeddings" as a git submodule. It provides a wrapper function to create sentence embeddings without performing the PCA step on the simple weighted average.

The Few-Shot-Text-Classification.ipynb notebook shows how to reproduce the accuracies obtained on the 20 Newsgroups dataset here.