eego

NLP experiments with ZuCo for COLING 2020

Usage on spaceml:
$ conda activate /mnt/ds3lab-scratch/noraho/anaconda3/envs/env-eego
$ CUDA_VISIBLE_DEVICES=7 python tune_combi_model.py

Set configuration to train or tune in config.py.

Feature extraction

ZuCo datasets available at: /mnt/ds3lab-scratch/noraho/coling2020

Word embeddings available at: /mnt/ds3lab-scratch/noraho/embeddings/glove-6B/ and /mnt/ds3lab-scratch/noraho/embeddings/bert/

NER

sequence to sequence word-level classification

Relation detection

multi-label sentence-level classification

Sentiment analysis

binary or ternary sentence-level classification only SR sentences from ZuCo 1