4 Commits

Author	SHA1	Message	Date
Yangqing Jia	deb1327b6e	Re-apply #266 ... Summary: Closes https://github.com/caffe2/caffe2/pull/404 Differential Revision: D4943280 Pulled By: Yangqing fbshipit-source-id: c0988598d8ccb8329feac88382686324b90d4d46	2017-04-25 21:17:04 -07:00
Yangqing Jia	a48062b1a2	temporarily fix sync script bugs changes by reverting partially https://github.com/caffe2/caffe2/pull/266/files	2017-04-24 15:49:22 -07:00
Aaron Markham	b93a7b134a	doxygen configs and updated python files to inc. doxygen tags (#266) ... * updated ubuntu instructions * updated ubuntu notes and troubleshooting * updated tutorials using local files * added doxygen python blocks for docs generation * doxygen related files for generating docs	2017-04-14 16:30:33 -07:00
Jiyan Yang	b401cb48fe	Make optimization methods configurable and allow flexible optimization settings ... Summary: This diff does the followings: 1. Add optimization options to model options in the UI for all workflows. 2. Allow different parameters to use different optimizers (or same optimizer with different settings, eg, learning rate). 3. Remove the default values for the `sparseDedupAggregator` field in the thrift file as the default value for that should just be `None` instead of 'sum'. 4. `fb/dper/layer_models/mlp_sparse.py` is deprecated. 5. Add calibration to two tower workflows. Reviewed By: kittipatv Differential Revision: D4767004 fbshipit-source-id: de92ea63fb0ff33f8581b1693479b723a68cd2d1	2017-04-01 23:02:21 -07:00