hammerlab / mhcflurry

75%

master: 73%

LAST BUILD ON BRANCH perf

branch: perf

CHANGE BRANCH

x

Reset

• perf
• 0.0.0
• 0.0.8
• 0.2.0
• 0.9.1
• 0.9.2
• 1.0.0
• 2017-11-13
• add-batch-normalization
• add-class1-ensemble
• add-class1-ensemble-rebased
• add-hyperparameters-base
• add-speed-script
• additions2
• antigen-presentation
• antigen-presentation-updates
• any-kmer-size
• bad-boy-keras
• before-sept2016-refactor
• bump-keras-version
• dataset
• docker-tweaks
• dockerize
• ensemble12
• expose-pretrain-decay
• fix-allelle-typo
• fix-commandline-predict-peptide-lengths
• fix-setup-for-py3
• fix-training-script
• get-supported
• hammer_tf_backend
• keras-layers-for-variable-length-sequences
• kubeface-model-training
• master
• move-imputation-from-experiments
• move-imputation-to-training-script
• moving-helpers-from-dataset-to-common
• multi-input-multi-output-models
• pan-allele-class1
• percentile-ranks
• pin-keras
• pypi0.9.2
• refactor1
• refactor1-rebased
• remove-experiements-and-notebooks
• remove-scripts
• remove-versioneer
• rewrite1
• rohan/prediction_speed
• script-updates
• small-fixes
• small-fixes-to-random-negative-samples
• small-fixes2
• split-py2-into-separate-docker
• timodonnell-patch-1
• use-imputation-in-training-script
• work0.9.3

Build # 558

Build Type

push

travis-ci

Committed by timodonnell

Commit Message

Reuse Keras models optimization

This should give a substantial prediction speed improvement as identified by @rohanpai

* Add borrow_cached_network() method to Class1NeuralNetwork
* Use cached models at in Class1NeuralNetwork.predict
* Set names for all neural network layers so the model JSON is identical for identical architectures

Run Details

1257 of 1671 relevant lines covered (75.22%)

3.01 hits per line