You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
Daniel Muckerman 60de0b4156 Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
datasets Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
models Initial commit 3 years ago
.gitignore Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
README.md Update README 3 years ago
dataset.py Initial commit 3 years ago
distill.py Initial commit 3 years ago
dump_pickles.py Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
intent_filters.py Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
intents.pkl Initial work on standalone query parsing 3 years ago
models.py Initial commit 3 years ago
num_words.pkl Initial work on standalone query parsing 3 years ago
prune.py Initial commit 3 years ago
requirements.txt Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
slot_filters.pkl Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
slots.pkl Initial work on standalone query parsing 3 years ago
snips_joint Initial work on standalone query parsing 3 years ago
test.py Initial commit 3 years ago
test_query.py Add slot filtering by intent to attempt to increase slot accuracy 3 years ago
timer.py Initial commit 3 years ago
train.py Initial commit 3 years ago
util.py Initial commit 3 years ago
word2idx.pkl Initial work on standalone query parsing 3 years ago
wordvecs.pkl Initial work on standalone query parsing 3 years ago

README.md

pruned-nlu

Requirements

  • >= Python 3.7
  • >= PyTorch 1.4.0
  • seqeval

Setup

Download GloVe embeddings:

wget http://nlp.stanford.edu/data/glove.6B.zip
unzip glove.6B.zip -d glove

Files

  • train.py, test.py, distill.py, timer.py, prune.py: runnable scripts, check each file's argparse for options and details
  • models.py: intent detection, slot-filling, and multi-task (joint intent detection and slot filling) CNN models
  • dataset.py: dataset loading abstractions
  • util.py: common code
  • models/: pretrained models, 5 duplicates of each
  • datasets/: prepared ATIS and Snips datasets

Contact

ojas@utexas.edu