60bb0e4e50
Uncommentable section in Dockerfile to autosave .py and .html alongside .ipynb
2017-12-11 16:19:24 +01:00
1d370f4001
nbdiff_checkpoint parameter parsing fixed
2017-12-11 06:52:17 +01:00
30fef69ed0
rm_empty_subdirs changed to require explicit argument
...
(defaulting to current dir withdrawn as potentially harmful)
2017-12-10 18:18:33 +00:00
2c9b0dab8a
Merge branch 'master' into ziembla-mod
2017-12-10 18:53:34 +01:00
5bb9d6d3df
help message for nbdiff_checkpoint
2017-12-10 18:38:25 +01:00
6e4004f16c
scripts for jupyter notebooks cleanup, bin subdir on path
2017-12-09 20:17:56 +01:00
63c1523528
Replace n_inputs with n_outputs, fixes #125
2017-12-07 18:57:30 -08:00
53c2133bc7
Merge pull request #133 from stefan-it/fix-margin-loss-formula
...
capsnet: fix margin loss formula
2017-12-07 18:27:02 -08:00
ddb9784176
tensorflow version unpined, tensorboard support, home variable fix
2017-12-04 12:08:22 +01:00
107de89304
Nbdime patching ignored if the original file was changed
2017-12-01 11:28:18 +01:00
e0b519c0a6
Merge branch 'master' into dev-woj
2017-12-01 11:04:29 +01:00
ef9df82689
Dockerfile publishable cleanup, git diff filter testing removed
2017-12-01 10:56:36 +01:00
72621ecdc5
Fix the link to the first video, and add a link and an embed for the second video
2017-11-30 19:00:21 +01:00
c50b5d9b3b
capsnet: fix margin loss formula
2017-11-30 17:36:19 +01:00
8586120c3d
Git filter testing demo
2017-11-30 12:59:26 +01:00
8d16b3061d
Patches to nbdiff for skipping noisy metadata, some local config
...
Nbdiff --ignore-details skils autoscroll, collapsed, deletable,
editable, toc (pull request on the way).
Enabling empty pass, no git pager, ignoring gitdiff nbdiff details.
2017-11-30 12:09:16 +01:00
4453715759
Merge branch 'master' into dev-woj
2017-11-30 06:28:52 +01:00
e7a1273c48
Docker environment minutiae
...
Docker compose project name set to avoid collisions, smiley dropped from README heading
2017-11-30 06:09:45 +01:00
6b166b3729
Fix a few errors in extra_capsnets.ipynb
2017-11-29 21:43:08 +01:00
340487cbef
Merge pull request #130 from ziembla/dev-woj
...
Docker environment enhancements
2017-11-29 21:28:40 +01:00
7093626bc2
Jupyter switched to token authentication by default, enabled nbdime integration in git diff
2017-11-29 16:20:07 +01:00
93623e8cc2
Dockerfile argument to set container user UID
...
The UID of created user should be set to the UID of the user that will
be running the image/container to set appropriate permissions on files
within notebooks directory. Default 1000 should be ok for users using
the default account created while installing linux.
2017-11-28 10:33:20 +01:00
fc355ca6b9
Rolled-back graphviz
...
I gave up on enabling DT visualization for the moment, as graphviz integration
stopped working... I can't run dot in the image any more, receiving:
Could not load "/opt/conda/lib/graphviz/libgvplugin_pango.so.6" - file not found
As far as I was able to diagnose thats (with ldd), that's because libiconv.so.2
is not available. But according to what I see in mailgroups that should not be
needed (on debian at least)...
2017-11-28 09:57:47 +01:00
4fa5beb93a
Docker environment enhancements
...
- rearranged a Dockerfile to allow for incremental build
- switched running from root to "default new user"
- added the (easy to opt-out) configuration to use blank password
- added python-graphviz which enables DT visualization in notebooks
- added nbdime for "sensible notebook comparison"
- added custom command to "nbdiff" a notebook with its checkpointed version
- added simple README.md
2017-11-27 17:16:51 +01:00
ca0f70a6b9
Fix latex math formatting for Jupyter.org's notebook viewer
2017-11-27 14:28:41 +01:00
e5401ac62d
Add capsule networks notebook
2017-11-27 14:21:28 +01:00
100e478315
Merge branch 'master' of github.com:ageron/handson-ml
2017-11-26 16:59:58 +01:00
eb4554ed72
Merge pull request #129 from Kxrr/master
...
Fix incorrect description in chapter 2
2017-11-24 17:32:39 +01:00
767affca86
Fix incorrect description in chapter 2
2017-11-25 00:02:08 +08:00
0cd0ed6bf9
Merge pull request #127 from StevenBunkley/dev
...
Created docker script
2017-11-22 09:17:35 +01:00
17bb3c4848
Remove image on stop
...
remove the image when you stop the container
2017-11-21 15:06:21 -06:00
ca1806aa1d
Created docker script
...
I created a docker script to install and run the project.
2017-11-21 13:24:43 -06:00
b17a6c8e63
Merge branch 'master' of github.com:ageron/handson-ml
2017-11-09 16:40:44 +01:00
8a6c7da0a9
Scale X_state down to [-1, 1] range in chapter 16
2017-11-09 16:35:44 +01:00
7686839b36
Use int8 instead of float64 to represent pixel values: divides RAM footprint by 8
2017-11-09 13:17:24 +01:00
2a02668e5e
Fix JSON formatting in new comment in the ch 11 notebook
2017-11-03 13:43:56 +01:00
694f35ee16
Merge pull request #113 from ChrisQlasty/master
...
Scikit-learn 19.0 updates on .fit( ) arguments
2017-11-03 13:38:48 +01:00
99b91ae989
Update 11_deep_learning.ipynb
2017-11-03 12:07:37 +01:00
e5fb7fb175
Comments fix
2017-11-03 12:04:54 +01:00
ef8f8a66fa
0.18 version left with 0.19 suggestions commented
...
The previous version of the .fit( ) with the suggestion of how it should be done from the version 0.21.
2017-11-03 12:00:28 +01:00
ca61fc15ab
Scikit-learn 19.0 updates on .fit( ) arguments up.
...
I have left unnecessary ',' at the end of the fit_params line, sorry for that.
2017-11-02 18:54:12 +01:00
ceaef6e7a1
Scikit-learn 19.0 updates on .fit( ) arguments
...
Adopting code to my needs I have found that in the scikit-learn 19.0 they recommend to put params list directly into the .fit( ) methods. That also makes the code more understandable for me as now it is more clear where these values go to (fit( ) function of DNNClassifier).
Hope this makes sense.
2017-11-02 18:30:30 +01:00
fd0ce384f2
Set DropoutWrapper's input_keep_prob parameter using a placeholder, fixes #111
2017-10-27 16:19:15 +02:00
b39d5366f7
Right angle is pi/2, not pi/4. One reason why tau > pi ;)
2017-10-27 13:03:15 +02:00
e09fa470ce
Merge branch 'master' of github.com:ageron/handson-ml
2017-10-27 12:58:29 +02:00
72a747ed74
Replace test set with validation set in code example from page 269
2017-10-27 12:58:21 +02:00
4a9b2fa6bc
Merge pull request #107 from akellyirl/patch-1
...
Update 15_autoencoders.ipynb
2017-10-25 14:57:09 +02:00
59ab30722e
Update 15_autoencoders.ipynb
...
PCA with a linear Autoencoder Section:
The calls to numpy.random require an import to work.
Added:
import numpy.random as rnd
2017-10-21 16:36:39 +01:00
7d7ccce9d3
Fix factorize() bug, due to housing_cat being a DataFrame, not a Series, fixes #104
2017-10-20 20:07:40 +02:00
da9ed522b9
Merge branch 'francotheengineer-master'
2017-10-18 09:45:16 +02:00
ziembla