Skip to content
Snippets Groups Projects

Repository graph

You can move around the graph by using the arrow keys.
Select Git revision
  • amir
  • gan
  • integrate_structures
  • lstm
  • master default protected
  • newlstm
  • oldmaster
  • predict
  • speed-up
  • tf20
  • towards_tf2
  • updates
  • v3.0.0
  • v2.0.0
  • v1.2.2b0
  • v1.2.1
  • v1.2.0
  • v1.1.1
  • v1.1.0
  • v1.0.4
  • v1.0.3
  • v1.0.2
  • v1.0.1
  • v1.0.0
  • 0.1.10
  • 0.1.9
  • 0.1.8
  • 0.1.7
  • 0.1.6
  • 0.1.5
  • 0.1.4
  • 0.1.3
32 results
Created with Raphaël 2.2.09Nov8632131Oct302926252423212019181614131211106532129Sep2827262523221915128764131Aug29282517427Jul2420171312108765429Jun282726231914231May302386523Apr2216724Mar2317161386424Feb2216156Jan4321Dec196130Nov2928242118171614111098543231Oct30282726252321181716141312111097626Sep252321129875230Aug2928252416121110Merge branch 'predict' into 'master'Merge branch 'master' into 'predict'Merge branch 'organizing-transfer-learning' into 'master'Renamed the keyword argumente is_training_mode to mode and use the tf.estimator.Modekeys to switch between training/validation/prediction modeslog_device_placement=TrueExist if all files already existcorrect syntaxcode clean-upfilter existing filesAdd base_archremove repeat. Use dataset.repeat insteadmake sure the biofiles are keptKeep regenerating labels and keysupdate train generic scriptAllow for transfer learning maybe?Convert the bio generator to a classRemoved debuging stuffOrganized the transfer learning variables for the estimatorsReorganized the keyword arguments for transfer learningUpdated tripletMerge branch 'hdf5-loading' into 'master'Removed the zerosMerge branch 'triplet-transfer-learning' into 'master'Merge branch 'master' of gitlab.idiap.ch:bob/bob.learn.tensorflow into hdf5-loadingImplemented transfer learning mechanism with the triplet estimatorFixed severe problem with the siamese estimator. Sorry to push this directly to the master, but this is an emergency. The estimator was super wrongCreated mechanism that loads data from bob's HDF5 and it converts to tensorRemoved import from the old analyserMerge branch 'batch-tfrecord' into 'master'Removing LFW tfrecord generatorUpdated the LFW to tfrecord script to the new feature formatCreated a function that batches a tf-record in order and apply data augmentationMerge branch 'predict' into 'master'Merge branch 'issue-44' into 'master'log and accuracyLog accuracy and lossAdd repeat option for training pipelinesloss is summarized by default in tensorflow 1.4load data is different nowDon't initialze the estimators