wellcometrust/grants_tagger

Issues

• Add extra data and references to grants in Argilla

  #264 opened a year ago by AndreiXYZ
  0

• Add script to upload data to Argilla for annotation

  #261 opened a year ago by AndreiXYZ
  0

• Add option to pre-annotate grants data

  #259 opened a year ago by AndreiXYZ
  0

• Create stratified sample for grants

  #257 opened a year ago by AndreiXYZ
  0

• Get evaluation to run with new model

  #255 opened a year ago by AndreiXYZ
  0

• Model predictions are erratic

  #250 opened 2 years ago by AndreiXYZ
  2

• Implement proper saving/loading of BertMesh to/from local

  #254 opened 2 years ago by AndreiXYZ
  0

• Decouple production code and experimental code/package

  #201 opened 2 years ago by aCampello
  3

• Make a recipe to push to datalabs-prod

  #161 opened 2 years ago by aCampello
  4

• Come up with xlinear docker container service

  #196 opened 2 years ago by aCampello
  1

• Experiment with increased n_gram and max_features for Tfidf transformation

  #149 opened 2 years ago by aCampello
  2

• Train XLinear on sagemaker

  #204 opened 2 years ago by aCampello
  1

• Move steps done in Travis CI into Github Actions

  #248 opened 2 years ago by AndreiXYZ
  1

• Try man negative sampling for XLinear

  #128 opened 2 years ago by nsorros
  1

• Try imbalanced ratio 0.2 for XLinear

  #130 opened 2 years ago by nsorros
  1

• Use own Mesh Tree instead of learned one for XLinear

  #144 opened 2 years ago by aCampello
  1

• Try only-top-k and beam size 20 for XLinear

  #131 opened 2 years ago by nsorros
  1

• Paper experiments 📜

  #188 opened 2 years ago by nsorros
  4

• Try tfn+man negative sampling for Xlinear

  #129 opened 2 years ago by nsorros
  1

• SHAP for XLinear

  #203 opened 2 years ago by aCampello
  1

• Switch to poetry and tox

  #251 opened 2 years ago by nsorros
  0

• Refactor tests

  #178 opened 3 years ago by nsorros
  4

• Separate get_grants in a different dvc.yaml

  #207 opened 3 years ago by aCampello
  0

• Per-label accuracy should live in s3?

  #210 opened 3 years ago by aCampello
  0

• Compute f1_score per label

  #150 opened 3 years ago by aCampello
  1

• Folder organisation

  #202 opened 3 years ago by aCampello
  0

• Fix deployment technical debt

  #198 opened 3 years ago by aCampello
  0

• Onlytop_k not working

  #193 opened 3 years ago by aCampello
  0

• Train default BertMesh

  #127 opened 3 years ago by nsorros
  0

• Dependency hell

  #185 opened 3 years ago by aCampello
  1

• Xmas shortlist 🎄

  #169 opened 3 years ago by nsorros
  3

• Add a way to test dvc repro on sample / test dataset

  #176 opened 3 years ago by nsorros
  2

• Remove torch and core dependency from WellcomeML

  #153 opened 3 years ago by nsorros
  3

• Fix train fail in tox

  #154 opened 3 years ago by nsorros
  1

• Stop using aws sync in favour of dvc pull

  #133 opened 3 years ago by nsorros
  3

• Don't import tensorflow on `grants_tagger.utils`

  #165 opened 3 years ago by aCampello
  0

• Train Xtransformer using PubMedBert for 10 epochs

  #125 opened 3 years ago by nsorros
  1

• sparse_y should live in MeshCNN and not CNNClassifier

  #134 opened 3 years ago by nsorros
  1

• Use tox for tests

  #141 opened 3 years ago by nsorros
  1

• Make default logger level to Error including loggers of libraries

  #155 opened 3 years ago by nsorros
  1

• Logger in tune threshold does not inherit from root

  #156 opened 3 years ago by nsorros
  0

• Train Xtransformer using PubMedBert with params inspired from paper. TBD.

  #126 opened 3 years ago by nsorros
  0

• Create literature document

  #132 opened 3 years ago by nsorros
  1

• Bump wellcomeml to 2.0.0

  #146 opened 3 years ago by aCampello
  1

• Grants tagger broken due to dependencies from WellcomeML

  #152 opened 3 years ago by nsorros
  1

• train command fails locally due to ConnectionError

  #142 opened 3 years ago by nsorros
  0

• Implement fit of ScienceEnsemble and use in DVC

  #135 opened 3 years ago by nsorros
  0

• Train Xtransformers using PubMedBert in years 2018,2019 and test in 2020,2021

  #123 opened 3 years ago by nsorros
  0

• Train Xtransformer using PubMedBert in 2016-2019

  #124 opened 3 years ago by nsorros
  0

• Train Xtransformers with PubMedBert and truncate in 256

  #122 opened 3 years ago by nsorros
  1