summaryrefslogtreecommitdiff
path: root/src/caffe/solver.cpp
AgeCommit message (Expand)AuthorFilesLines
2017-01-23Restore can be invoked on rank > 0Cyprien Noel1-1/+0
2017-01-06Switched multi-GPU to NCCLCyprien Noel1-25/+19
2016-11-25Revert "solver: check and set type to reconcile class and proto"Evan Shelhamer1-12/+0
2016-11-21solver: check and set type to reconcile class and protoEvan Shelhamer1-0/+12
2016-02-25drop Net inputs + Forward with bottomsEvan Shelhamer1-5/+3
2016-01-26Merge pull request #3132 from bwilbertz/LastIterationLossJeff Donahue1-13/+24
2015-11-26replace snprintf with a C++98 equivalentTea1-5/+3
2015-11-12Fix loss of last iteration when average_loss > 1Benedikt Wilbertz1-13/+24
2015-10-16Add automatic upgrade for solver typeRonghang Hu1-1/+1
2015-10-16Split solver code into one file per solver classRonghang Hu1-811/+0
2015-09-18refine format of switch case in solverYan Chen1-8/+8
2015-09-14Merge pull request #3049 from seanbell/check-snapshot-prefixRonghang Hu1-0/+19
2015-09-14Check that the snapshot directory is writeable before starting trainingSean Bell1-0/+19
2015-09-09Minor: missing space in string formattingSean Bell1-1/+1
2015-08-22Add signal handler and early exit/snapshot to Solver.J Yegerlehner1-8/+62
2015-08-14Adam solverPatWie1-0/+104
2015-08-13Merge pull request #2903 from ronghanghu/multi_gpuRonghang Hu1-21/+55
2015-08-12Data Layers Parallel for Multi-GPURonghang Hu1-6/+17
2015-08-10Clean up and modernize AdaDelta code; add learning rate support; add addition...Matthias Plappert1-156/+118
2015-08-10Updated AdaDelta for modern Caffe; reduced iterations on multi-iter testsKevin Bache1-23/+9
2015-08-10Implement AdaDelta; add test cases; add mnist examplesMohamed Omran1-0/+199
2015-08-09Multi-GPUCyprien Noel1-17/+40
2015-08-09Use net_->learnable_params() instead of net_->params() in RMSpropRonghang Hu1-1/+1
2015-08-08Implement RMSProp SolverEren Golge1-0/+76
2015-08-07Net: add learnable_params_ used by solvers to correctly handle shared paramsJeff Donahue1-32/+12
2015-08-07Snapshot model weights/solver state to HDF5 files.Eric Tzeng1-30/+134
2015-05-28directly normalize accumulated gradientsEvan Shelhamer1-3/+29
2015-05-27adjust local learning rate and decay according to gradient accumulationEvan Shelhamer1-2/+3
2015-05-27zero-init param diffs in gradient checkerJonathan L Long1-1/+1
2015-05-27zero-init param diffs and accumulate gradientsJonathan L Long1-1/+26
2015-05-27Solver::MakeUpdate() -> Solver::ApplyUpdateEvan Shelhamer1-2/+2
2015-05-26deduplicate decay and local rate in solver updatesEvan Shelhamer1-18/+5
2015-05-26Refactor solvers regularization and logging codeCyprien Noel1-293/+207
2015-05-14Merge pull request #2072 from jeffdonahue/final-snapshot-off-by-oneJon Long1-5/+8
2015-03-13Fix for solver issue pointed out by @moskewcz in #1972Jeff Donahue1-1/+1
2015-03-08Increment iter_ before snapshotting, remove +1 logic -- fixes finalJeff Donahue1-5/+8
2015-03-03Blobs are ND arrays (for N not necessarily equals 4).Jeff Donahue1-10/+4
2015-02-17no phase for the solver to orchestrateEvan Shelhamer1-4/+0
2015-02-13Add gradient clipping -- limit L2 norm of parameter gradientsJeff Donahue1-0/+27
2015-01-29Enhancements for debug_info to display more information.Jeff Donahue1-0/+1
2015-01-09improve const-ness of NetJonathan L Long1-10/+13
2015-01-02break out Step from SolverJonathan L Long1-35/+38
2014-10-17back-mergeSergey Karayev1-1/+1
2014-10-16Fix init current_stepSergio1-0/+1
2014-10-10Added Multistep, Poly and Sigmoid learning rate decay policiesSergio1-6/+31
2014-10-06correct naming in comment and message about average_lossJonathan L Long1-1/+1
2014-10-06[fix] check solver prototxt parsingJonathan L Long1-1/+1
2014-10-01Back-merge branch 'master' into dev (for fix in PR #1203)Jeff Donahue1-4/+4
2014-10-01[fix] solver indexing of output blobs was incorrect for non-singletonJeff Donahue1-4/+4
2014-09-18fix warningqipeng1-1/+1
generated by cgit v1.2.3 (git 2.25.1) at 2024-04-20 11:19:53 +0000