Skip to content
Snippets Groups Projects
Select Git revision
  • dev
  • dev-train-models
  • main default protected
  • model-standard-data
4 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.029Nov242331Oct27262524111022Aug21151020Jul191311107630Jun2922211412210MayAdd space between headline and bodymodel-standard-…model-standard-dataAdd upsamplingChange average of f1 scoreMerge branch 'model-standard-data' of https://sissource.ethz.ch/oschmanf/ppg-moderation-classifier into model-standard-dataAdapt metrics to only f1Add option to make out of sample splitsAdd option to make out of sample splitsSave run historyAdd f1 score during trainingAdapt training for new datasetMerge branch 'model-standard-data' of https://sissource.ethz.ch/oschmanf/ppg-moderation-classifier into model-standard-dataChange paramsAdapt code to split csv and parquet datasetsAdapt code to allow training with Standard dataMerge branch 'dev'mainmainAllow to use list of ranges for hsprobdev dev-train-m…dev dev-train-modelsAdd params for local runSmall changed during training (e.g. number of epochs)Ignore saved modelsAdd scripts to run training for IMDBUpdate of bash scriptsMerge branch 'dev-train-models' into 'main'Add parameters for evaluation of pretrained modelsAdd filtering by hsprobAdd bash script for train runs on EulerAdd name of pretrained model as command line argument and to logsUpdate of README.mdAdd command line arguments for newspaper and topicGrouping of util functions for training and evaluation in separate filesGrouping of util functions for training and evaluation in separate filesFix difference between remote and local codeAdjust averaging of precision, recall, f1Adjust averaging of precision, recall, f1Only return most common topicsFix issue with F1 computingSmaller changes in train scriptsSpeed up inference by loading tokenizer and model seperately and not as pipelineAdd more evaluation metrics to BERT evaluationDrop rows with row indices when removing duplicatesAdjust params for training