search

sIncerass / prag_generation

[NAACL 2019] code for "Pragmatically Informative Text Generation" https://arxiv.org/abs/1904.01301
47 stars 9 forks source link

Unable to find measure_scores.py #1

Closed dfayzur closed 5 years ago

dfayzur commented 5 years ago

Hello,

Thank you very much for the code. I was trying to run the code. But there is a file (measure_scores.py) missing. Would you please add the file to the repo.

sIncerass commented 5 years ago

Sure, I upload the "e2e-metrics" directory. To use the "measure_score.py", you can run as

"python2 measure_scores.py ../data/devset.csv.multi-ref ../experiments/select_model_v1/weights.epoch8.devset.predictions.txt_incre_0.2_new"

sIncerass commented 5 years ago

To reproduce my experiment, you can just use the model in the "experiment" directory since we only change the inference process.

dfayzur commented 5 years ago

Thank you very much for adding measure_scores.py in the repo. The training and prediction is working without any issues. Closing it.