Age | Commit message (Collapse) | Author | Files | Lines | |
---|---|---|---|---|---|
2018-01-31 | v0.5.0 | huifang | 20 | -1113/+869 | |
2017-10-20 | 1. support ACL v17.10 | honggui | 2 | -11/+7 | |
2. the performance of pooling was increased 30X 3. Removing CORASE timer | |||||
2017-10-11 | revised an compatibility issue with the new ACL 17.09 version. Documents ↵ | honggui | 3 | -4/+5 | |
will be updated later on | |||||
2017-10-10 | ACL concat function is not stable enough yet. Bypass it till the issue was ↵ | honggui | 1 | -1/+2 | |
fixed. | |||||
2017-08-26 | add support acl batch normal,direct conv, local connect, concat layers | honggui | 16 | -71/+807 | |
2017-07-02 | Issue: When OpenCL initialization fails, even if Caffe uses CPU-mode, it ↵ | Yao Honggui | 1 | -1/+11 | |
doesn't work properly. Solution: when OpenCL initialization fails, if Caffe uses CPU-mode, it can work properly in neon mode | |||||
2017-07-01 | To match the new ARM Compute Library version 17.06 | Yao Honggui | 1 | -1/+2 | |
2017-06-02 | 1. Porting Caffe onto ARM Compute Library. | Yao Honggui | 29 | -3/+1664 | |
2. The release version is 0.2.0 | |||||
2017-03-22 | Clarify batch norm parameter documentation. | Jonathan R. Williford | 1 | -4/+14 | |
2017-03-21 | sane h5df file type check for weights | max argus | 1 | -2/+1 | |
2017-03-07 | Merge pull request #4630 from BlGene/load_hdf5_fix | Evan Shelhamer | 4 | -12/+37 | |
Made load_hd5 check blob dims by default, instead of reshaping. | |||||
2017-02-16 | Merge pull request #4609 from intelfx/BVLC-work-buildsystem | Evan Shelhamer | 3 | -5/+16 | |
[cmake] Improvements to the build system | |||||
2017-02-16 | Merge pull request #5210 from ftokarev/patches | Evan Shelhamer | 1 | -2/+1 | |
Obsolete reference to `bool solver` in caffe.proto | |||||
2017-01-23 | Restore can be invoked on rank > 0 | Cyprien Noel | 1 | -1/+0 | |
2017-01-21 | Update a comment in caffe.proto | Fyodor Tokarev | 1 | -2/+1 | |
2017-01-19 | Merge pull request #5184 from shaibagon/fix_batch_norm_param_upgrade | Evan Shelhamer | 1 | -1/+7 | |
fix harsh upgrade_proto for `"BatchNorm"` layer to zero instead of clear params | |||||
2017-01-19 | Merge pull request #3365 from BonsaiAI/wrap-declarations-in-switch | Evan Shelhamer | 1 | -2/+2 | |
Fix syntax inside switch for Windows compilation | |||||
2017-01-18 | remove redundant operations in Crop layer (#5138) | xmyqsh | 2 | -35/+27 | |
2017-01-18 | Fix various documentation typos (#4172) | Adam Browne | 1 | -1/+1 | |
* fix typo (standaraized->standardized) * fix typo (convet->convert, etc..) * fix typo (incompartible->incompatible) * fix typo (does't->doesn't) * fix typo (decoded->decode) | |||||
2017-01-15 | fixing upgrade_proto for BatchNorm layer: be more conservative leave "name" ↵ | shai | 1 | -1/+7 | |
in param, only set lr_mult and decay_mult to zero | |||||
2017-01-06 | Using default from proto for prefetch | Cyprien Noel | 1 | -2/+1 | |
2017-01-06 | Switched multi-GPU to NCCL | Cyprien Noel | 26 | -635/+589 | |
2016-12-30 | Typos in test_inner_product_layer.cpp | Fyodor Tokarev | 1 | -5/+5 | |
2016-11-28 | Add the missing period | Zhou Mo | 1 | -3/+3 | |
2016-11-28 | Make lint happy (> 80 characters) | Zhou Mo | 1 | -4/+4 | |
2016-11-27 | fix many typos by using codespell | Zhou Mo | 8 | -11/+11 | |
2016-11-25 | Revert "solver: check and set type to reconcile class and proto" | Evan Shelhamer | 2 | -17/+0 | |
as pointed out by #5028 this does not achieve what it intended, and furthermore causes trouble with direct solver instantiation. revert commit e52451de914312b80a83459cb160c2f72a5b4fea | |||||
2016-11-22 | Merge pull request #5009 from shelhamer/solver-type-check | Evan Shelhamer | 2 | -0/+17 | |
Solver: check and set type to reconcile class and proto type | |||||
2016-11-22 | Merge pull request #4998 from chenzeyuczy/master | Jeff Donahue | 1 | -2/+2 | |
Add missing spaces besides equal signs in batch_norm_layer.cpp | |||||
2016-11-21 | solver: check and set type to reconcile class and proto | Evan Shelhamer | 2 | -0/+17 | |
the solver checks its proto type (SolverParameter.type) on instantiation: - if the proto type is unspecified it's set according to the class type `Solver::type()` - if the proto type and class type conflict, the solver dies loudly this helps avoid accidental instantiation of a different solver type than intended when the solver def and class differ. guaranteed type information in the SolverParameter will simplify multi-solver coordination too. | |||||
2016-11-18 | Merge pull request #4703 from cypof/avoid_missing_returns | Evan Shelhamer | 1 | -0/+7 | |
Avoids missing return values during build | |||||
2016-11-18 | Add missing spaces besides equal signs in batch_norm_layer.cpp | chenzy | 1 | -2/+2 | |
2016-11-16 | sigmoid cross-entropy loss: normalize loss by different schemes | Evan Shelhamer | 3 | -30/+91 | |
sig-ce loss handles all the same normalizations as the softmax loss; refer to #3296 for more detail. this preserves the default normalization for sig-ce loss: batch size. | |||||
2016-11-15 | sigmoid cross-entropy loss: ignore selected targets by `ignore_label` | Evan Shelhamer | 3 | -0/+70 | |
sig-ce learns to ignore by zeroing out the loss/diff at targets equal to the configured `ignore_label`. n.b. as of now the loss/diff are not properly normalized when there are ignored targets. sig-ce loss should adopt the same normalization options as softmax loss. | |||||
2016-11-01 | add the missing star in comment | nihui | 1 | -1/+1 | |
a trival commit which adds the missing star ;) | |||||
2016-10-27 | sigmoid cross-entropy loss: add GPU forward for full GPU mode | Evan Shelhamer | 2 | -3/+35 | |
close #3004 | |||||
2016-10-22 | Fix: made load_hd5 check blob dims by default. | max argus | 4 | -12/+37 | |
Size checks are needed for loading parameters to avoid strange bugs when loading data we continue to reshape. | |||||
2016-09-29 | slightly relax batch norm check | Benedikt Wilbertz | 1 | -4/+8 | |
2016-09-20 | Merge pull request #4600 from bwilbertz/fix_scale_layer | Jeff Donahue | 1 | -3/+11 | |
fix layerSetUp of scale_layer to not add bias blob when already present | |||||
2016-09-12 | batch norm: auto-upgrade old layer definitions w/ param messages | Evan Shelhamer | 1 | -1/+33 | |
automatically strip old batch norm layer definitions including `param` messages. the batch norm layer used to require manually masking its state from the solver by setting `param { lr_mult: 0 }` messages for each of its statistics. this is now handled automatically by the layer. | |||||
2016-09-12 | batch norm: hide statistics from solver, simplifying layer definition | Evan Shelhamer | 1 | -0/+8 | |
batch norm statistics are not learnable parameters subject to solver updates, so they must be shielded from the solver. `BatchNorm` layer now masks its statistics for itself by zeroing parameter learning rates instead of relying on the layer definition. n.b. declaring `param`s for batch norm layers is no longer allowed. | |||||
2016-09-09 | Avoids missing return values during build. | Cyprien Noel | 1 | -0/+7 | |
2016-09-09 | Benchmarking should not impact perf until timer is read | Cyprien Noel | 1 | -1/+2 | |
2016-09-01 | fix layerSetUp of scale_layer to not add bias blob when already present | Benedikt Wilbertz | 1 | -3/+11 | |
2016-08-31 | cmake: add option to link with OpenMP | Ivan Shapovalov | 1 | -0/+3 | |
Despite Caffe itself does not use OpenMP, explicitly linking to OpenMP should be done when one statically links to a BLAS library which uses OpenMP internally and does not provide proper CMake imported targets with proper dependencies (nobody this so far). | |||||
2016-08-31 | net.cpp: do not include test/test_caffe_main.hpp | Ivan Shapovalov | 1 | -2/+0 | |
2016-08-31 | cmake: refactor deps detection, specify all dependencies in the exported ↵ | Ivan Shapovalov | 2 | -3/+13 | |
caffe target This is the first step towards "modern" IMPORTED-targets-only CMake setup. The find_package modules still need to be rewritten and upstreamed in form of config exports where possible. | |||||
2016-08-11 | Merge pull request #4379 from alicanb/master | Jeff Donahue | 1 | -1/+1 | |
add default value to rms_decay and fix documentation | |||||
2016-08-06 | num in blob is deprecated | Fisher Yu | 1 | -2/+2 | |
2016-07-11 | Merge pull request #4359 from ibmsoe/fix-EmbedLayerTest | Evan Shelhamer | 1 | -1/+1 | |
fix equality check in EmbedLayer test |