e8d8ae0eab
add ELMo embeddings
2018-12-05 17:07:07 -08:00
17c09d977b
Add differentiable BLEU loss
...
-Using differentiable BLEU loss instead of cross_entropy loss
-it helps decreasing train-test evaluation gap
2018-11-27 15:22:38 -08:00
ca5818dcf1
Getting the best from both sides
...
The branches are getting messy! This branch is a clean updated version of the codebase which is actually compatible with 1) new version of pytorch 2) Almond project
2018-11-07 15:06:41 -08:00
f835558916
Reducing train_batch_tokens; 10k OOMs on 0.4.1
2018-10-25 16:02:48 -07:00
38ce8ac161
mv save out of val; add arg for no tensorboard logging
2018-10-25 16:02:48 -07:00
2bcd810957
update devices
2018-10-25 16:02:48 -07:00
09cebc62a3
enumerating predictions; exp with intermediate cove
2018-09-04 15:43:27 +00:00
8e85a13c87
Contextualized Word Vectors (CoVe; McCann et al. 2017)
2018-08-28 03:14:41 +00:00
97181e99e9
default gpus
2018-06-27 21:38:53 +00:00
e0e04702cc
rm dependency on .git; multi-gpu WIP; fine-grained time
2018-06-27 21:18:45 +00:00
fbf94815c8
replace runtime with decaNLP
2018-06-24 14:40:52 +00:00
ca1718b33c
MQAN default model
2018-06-22 17:22:10 +00:00
8997039aec
Initial commit
2018-06-20 12:36:21 +00:00
mehrad