Giters
kudkudak
/
common-sense-prediction
Common sense prediction using DL.
Geek Repo:
Geek Repo
Github PK Tool:
Github PK Tool
Stargazers:
4
Watchers:
6
Issues:
66
Forks:
0
kudkudak/common-sense-prediction Issues
Li filters out of vocab triplets from wiki, I don't
Closed
5 years ago
Comments count
2
Explore regularizers for glove
Closed
5 years ago
Comments count
1
Improvements to train scripts
Closed
5 years ago
Fixes to experiments
Closed
5 years ago
Compile Appendix with novel and trivial triplets from 4.1 section.
Closed
5 years ago
Rethink reporting F1 on imbalance of negative class bucket
Closed
5 years ago
vocab should be property of dataset, not embedding.
Closed
5 years ago
Comments count
2
Check relation of conceptnet5 to version used in paper
Closed
5 years ago
Grid to pick hyperparameter for each model
Closed
5 years ago
Comments count
2
Unit test human evaluation
Closed
5 years ago
Comments count
3
Re-evaluate one failed model on wiki buckets
Closed
6 years ago
Rescore on whole wiki baselines vs Li et al.
Closed
6 years ago
Think about balance of classes
Closed
6 years ago
Comments count
1
Prepare data for Dima
Closed
6 years ago
Interhuman agreement of scores
Closed
6 years ago
Comments count
1
Score 900 triplets forming base of wiki evaluation
Closed
6 years ago
Bucket wiki into 3 buckets - evaluation scripts
Closed
6 years ago
Fast computing distance script
Closed
6 years ago
Comments count
1
Dev re-make
Closed
6 years ago
Double check negative sampling
Closed
6 years ago
Human novelty evaluation
Closed
6 years ago
Write sketch of text
Closed
6 years ago
merge LiACLDatasetFromFile and LiACLSplitDataset
Closed
6 years ago
Improve human evaluation script
Closed
6 years ago
Comments count
1
Code script to add distances, run for wiki and conceptnet, analyze
Closed
7 years ago
Comments count
8
Label negative samples from test
Closed
7 years ago
Explore wiki distances
Closed
7 years ago
Examine regularization for glove embeddings when used together with OMCS
Closed
7 years ago
Comments count
1
Negative samples outside of batch
Closed
7 years ago
Comments count
4
Manually label the 5k conceptnet
Closed
7 years ago
Comments count
3
Human examination of split quality
Closed
7 years ago
Comments count
1
Check if distances on conceptnet extrinsic test are useful
Closed
7 years ago
Negative samples using ArgSim - human based examination
Closed
7 years ago
Run more models on new negative sample split
Closed
7 years ago
Small bug in argsim_threshold fnc: avg should count non-zeros
Closed
7 years ago
Comments count
1
Investigate whether we can use shitty triplets from CN as negative examples
Closed
7 years ago
Comments count
1
Create a golden set of "trivial/novel" pairs
Closed
7 years ago
Comments count
4
MaxSim with Fuel
Closed
7 years ago
Human based negative samples - decide on protocol
Closed
7 years ago
Perform human evaluation on wiki
Closed
7 years ago
Comments count
2
MaxSim - negative sampling using MaxSim
Closed
7 years ago
Groundword: do hp optimization for current DNN_CE and Factorized
Closed
7 years ago
Comments count
1
Try argsim and prototypical negative samples - effect on extrinsic and test gen
Closed
7 years ago
Comments count
1
Try to merge ArgSim and MaxSim
Closed
7 years ago
Closest-neighbour analysis
Closed
7 years ago
Add conceptnet extrinsic eval
Closed
7 years ago
Try to figure out why their model works well on extrinsic eval
Closed
7 years ago
Comments count
1
Evaluate on wiki
Closed
7 years ago
vegab improve meta.json
Closed
7 years ago
Add good configs for Factorized to config file
Closed
7 years ago
Previous
Next