wellcometrust/grants_tagger

Issues

• Train Xtransformer using PubMedBert

  #121 opened 3 years ago
  0

• Train default Xtransformers

  #120 opened 3 years ago
  0

• Save models and data under approach, dataset subfolder

  #119 opened 3 years ago
  0

• Run streamlit with XLinear

  #118 opened 3 years ago
  1

• Create CLI in cli.py and in each script

  #117 opened 3 years ago
  1

• Split train / test in separate script

  #115 opened 3 years ago
  1

• Save sparse weights in mesh tfidf-svm

  #114 opened 3 years ago
  0

• Return sparse probs in mesh models

  #113 opened 3 years ago
  0

• Optimise threshold for MeshXLinear

  #112 opened 3 years ago
  1

• Investigate whether there is a way to use sparse target for multilabel classification in tensorflow

  #111 opened 2 years ago
  1

• Release XLinear and align model versioning with package

  #110 opened 3 years ago
  0

• Add year in meta in grants_tagger and add filter by year param to preprocess_mesh

  #109 opened 3 years ago
  0

• Update README with PyPi instructions and Quickstart guide

  #107 opened 3 years ago
  0

• Rename results in docs to science_results

  #106 opened 3 years ago
  0

• Add AmazonCat13 to grants_tagger and recording results using CNNCassifier

  #105 opened 2 years ago
  1

• Improve predict / evaluate

  #104 opened 3 years ago
  0

• Add pull request template

  #102 opened 3 years ago
  0

• Switch to main from master

  #101 opened 3 years ago
  0

• Make grants tagger light to install

  #99 opened 3 years ago
  3

• Evaluate tf-idf-svm is pointing at the wrong label binarizer

  #97 opened 3 years ago
  0

• Add "time to train" as a report metric/document it

  #96 opened 3 years ago
  1

• Document which models/code version are in the production pipeline

  #95 opened 3 years ago
  0

• Check for data drift

  #94 opened 3 years ago
  0

• Update streamlit app so it uses 1.0

  #93 opened 3 years ago
  2

• Create models folder in grants_tagger and separate each model to separate files

  #83 opened 3 years ago
  0

• Train should only train not evaluate

  #82 opened 3 years ago
  0

• Split train / test before train and use test in evaluate

  #81 opened 3 years ago
  1

• Eliminate separate label_binarizer_path

  #80 opened 3 years ago
  2

• Fix sagemaker logs feedback when training

  #79 opened 3 years ago
  1

• Make mesh or all models public

  #75 opened 3 years ago
  0

• KerasVectorizer trained with all data in incremental learning which can cause errors

  #59 opened 3 years ago
  1

• Implement download function to download training data from EPMC

  #54 opened 3 years ago
  0

• Add docstrings and generate read the docs files

  #53 opened 4 years ago
  0

• Create streamlit app to showcase Mesh tagging

  #52 opened 3 years ago
  1

• Add to PyPi

  #51 opened 3 years ago
  1

• Pin requirements for reproducible builds

  #47 opened 3 years ago
  1

• Threshold is str instead of float in evaluate model

  #43 opened 4 years ago
  0

• y_batch_size fails when not provided

  #34 opened 4 years ago
  0

• Config 2020.04.2 and 2020.04.3 fail in save stage

  #33 opened 4 years ago
  0

• Save throws error for CNN and BiLSTM

  #32 opened 4 years ago
  0

• Create a model class for the baseline TFIDF-SVM model of mesh

  #24 opened 4 years ago
  0

• Use WellcomeVotingClassifier for science tagging

  #23 opened 4 years ago
  0

• Create entrypoint for grants_tagger

  #22 opened 4 years ago
  0

• Add tests for predict

  #19 opened 4 years ago
  0

• Refactor run_config and run_mesh_config

  #18 opened 4 years ago
  0

• Access to this repo

  #15 opened 4 years ago
  2

• Dependency conflict in version 2020.7.0

  #13 opened 4 years ago
  0

• Include models in packaged wheels

  #8 opened 4 years ago
  0

• Add an easy way to add custom preprocessing

  #4 opened 3 years ago
  1

• Differentiate syncing for science tags vs mesh

  #2 opened 4 years ago
  1