Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

merged master from main repo into deconv-deep-vis-toolbox branch #2

Open
wants to merge 89 commits into
base: deconv-deep-vis-toolbox
Choose a base branch
from

Conversation

asanakoy
Copy link

Updated branch with latest commits from https://github.com/BVLC/caffe master .
Tested on compatibility with https://github.com/yosinski/deep-visualization-toolbox master branch.

lukeyeager and others added 30 commits February 29, 2016 12:28
MKLROOT variable is set by MKL scripts, so it also should be used in Makefile.
[bulid] correct MKL_DIR to MKLROOT
…tiation-fix

Use lazy initialization to reuse ordered dict/list creations to save…
Set propagate_down=true to force backprop to a particular bottom
Issue with the exp layer when the base is e
The latest versions of protobuf do not reveal empty message fields with dir(). This uses the documented way of determining all of a message's fields and so is compatible with past and future versions of protobuf.
[docs] typo in help text for "-model" option
Latest CUDA versions are all compatible, and
Caffe has been compatible with cuDNN v4 since PR BVLC#3439
Caffe runs on Ubuntu, OS X, and RHEL (+ company) in master
with branches for OpenCL and Windows.

Docker is a nice route to out-of-the-box brewing.
bring install docs up-to-date and better highlight included details
check only the dimensions to be cropped for compatible sizes and offsets
Fix Crop layer dimension checking to only check cropped dimensions
[docs] typo in docs/installation.md
The -weights flag is somewhat easy to miss as it's only in one command,
but is the crucial thing that anyone searching for 'how to finetune' is
looking for. Hopefully this more clearly points out the '-weights' flag,
which might otherwise be overlooked in this tutorial.
…ult-label

Don't force datum.label=0 in array_to_datum
…torial

[docs] explicitly point out -weights flag in fine-tuning tutorial
Instead, double the map size on the MDB_MAP_FULL exception.
shelhamer and others added 30 commits May 4, 2016 13:47
Note that `make clean` clears build and distribute dirs
Also add a test that reshapes a blob to shape (0, 5).
Add parameter layer for learning any bottom
[bug] Catch MDB_MAP_FULL errors from mdb_txn_commit
[build] CMake customisable version/soversion
a comment misses a space char
cuDNN v4 is still supported.
Add cuDNN v5 support, drop cuDNN v3 support
Window data foreground/background nonzero check
Fix Makefile CUDA_VERSION extraction on OSX Yosemite
fix problems in net_surgery.ipynb
handle spaces in image file names

Thanks @bchu for an earlier implementation.
* Run on Ubuntu 14.04
* Test cuDNN builds
* Build with OpenBLAS

NOTE: Python3 build only works with CMake
…ests

Remove misleading comment from a test file
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.