Commit Graph

79 Commits

Author SHA1 Message Date
Reuben Morais
75091a8ef3 Switch prod model to a streaming based one 2018-08-02 23:46:13 -03:00
Reuben Morais
c579b74a1d Switch to updated head in TensorFlow r1.6 branch 2018-08-02 13:25:24 -03:00
Reuben Morais
ac41fe0006 Disable AOT tasks 2018-08-02 13:22:24 -03:00
Reuben Morais
cc84bb658a Remove some deepspeech_utils references that crept up during rebase 2018-08-02 13:22:24 -03:00
Reuben Morais
400c9c0e5d Streaming inference 2018-08-02 13:22:24 -03:00
Alexandre Lissy
e809e4746a Remove deprecated tests
We are now only relying on upstream TensorFlow packages and aiming at
being compatible. Makes no sense to continue supporting those,
especially since they are disabled.
2018-07-26 12:02:59 +02:00
Alexandre Lissy
298d11cd76 Add test coverage for training from frozen model 2018-07-24 13:02:53 +02:00
Alexandre Lissy
b455b24236 Remove useless training dep 2018-07-19 15:53:39 +02:00
Alexandre Lissy
d276bcab1f Force NodeJS installation from NodeSource repo
Fixes #1450
2018-07-09 23:07:14 +02:00
Alexandre Lissy
ab2f150d4a Switch to Python 3.7.0 stable
Fixes #1435
2018-07-03 10:08:05 +02:00
Alexandre Lissy
d08ef842bc Add Python 3.7.0b5
Fixes #1395
2018-06-05 22:21:24 +02:00
Alexandre Lissy
ad6a0704a7 Add NodeJS v10.x
Fixes #1394
2018-06-05 20:27:10 +02:00
Alexandre Lissy
05bc98ca86 Also remove aarch64 from packages pulling 2018-04-28 18:43:19 +02:00
Alexandre Lissy
5ee4c4a0d2 Remove Aarch64 packages from upload 2018-04-28 15:23:23 +02:00
Alexandre Lissy
5520a6b29d Improve tag handling 2018-04-28 02:13:13 +02:00
Alexandre Lissy
f4491531cf Use new 'tag' event
X-DeepSpeech: NOBUILD
2018-04-28 01:14:55 +02:00
Alexandre Lissy
258a444364 Add missing branch filtering for v0.2.0-alpha.0 2018-04-28 00:42:37 +02:00
Alexandre Lissy
e01784894e Unify version definition
Fixes #1293
2018-04-27 09:44:51 +02:00
Alexandre Lissy
6f0751d27d Switch to new TensorFlow r1.6 artifacts 2018-04-19 22:03:02 +02:00
Alexandre Lissy
9a92823db2 Add aarch64 python wheel build 2018-04-19 12:48:31 +02:00
Alexandre Lissy
4e53683c43 Add ARM64 builds 2018-04-19 11:36:53 +02:00
Alexandre Lissy
8382791536 Fix jessie to stretch naming 2018-04-18 16:07:40 +02:00
Alexandre Lissy
021de3c2a7 Enable C++, NodeJS tests on RPi3
Fixes #1303
2018-04-16 11:35:54 +02:00
Alexandre Lissy
342f16e5bf Update tasks names to ARMv7 2018-04-16 11:35:54 +02:00
Alexandre Lissy
f30bb0e5a2 Update to TensorFlow r1.6 with PYTHON_BIN_PATH forced 2018-04-15 16:30:15 +02:00
Alexandre Lissy
f77a5cdd2a Switch to GCC 7.2 from Linaro and Raspbian Stretch
Fixes #1304
2018-03-19 20:28:06 +01:00
Alexandre Lissy
6c4d4f1e68 Expose versions numbers on stderr
Fixes #1292
2018-03-16 09:46:31 +01:00
Alexandre Lissy
bb1fcc0542 Increase pip default timeout
Fixes #1289
2018-03-15 09:39:09 +01:00
Alexandre Lissy
0a9bfbd41f Switch to TensorFlow r1.6 2018-03-12 23:01:28 +01:00
Alexandre Lissy
164f67c972 Add Python2.7 ucs2+ucs4 variants
Fixes #1276
2018-03-01 17:51:52 +01:00
Alexandre Lissy
62034f7bf6 Switch to use Amazon EC2 mirrors
Fixes #1250
2018-02-15 17:38:58 +01:00
Alexandre Lissy
e4773028ba Update doc for mmap-able model
Fixes #1248
2018-02-15 17:23:33 +01:00
Alexandre Lissy
d7653d749b Use LDC93S1 small language model
How to regenerate:
 - Get a KenLM build
 - $ (tr -d '[:digit:]|[:punct:]' < LDC93S1.txt | tr '[:upper:]' '[:lower:]'; head -n 500 ../lm/vocab.txt) > vocab.txt
 - $ bin/lmplz --prune 0 0 1 --order 5 --text vocab.txt --arpa vocab.pruned.arpa
 - $ bin/build_binary  -s vocab.pruned.arpa vocab.pruned.lm
 - $ generate_trie ../alphabet.txt vocab.pruned.lm vocab.txt vocab.trie

Fixes #1245
2018-02-15 14:56:02 +01:00
Alexandre Lissy
0c323e8c67 Support MemmappedFileSystem
Fixes #1231
2018-02-15 12:18:52 +01:00
Alexandre Lissy
bcfd4fd409 Update for r1.5 2018-02-14 11:58:48 +01:00
Alexandre Lissy
eac987a09c Switch to reference TensorFlow r1.4 branch 2018-02-09 19:03:32 +01:00
Alexandre Lissy
72298b8f6d Add different samplerate samples
Fixes #1022
2018-02-08 20:48:53 +01:00
Alexandre Lissy
11fcae4bc7 Proper re-use of Bazel cache
Fixes #1189
2018-01-31 18:50:36 +01:00
Alexandre Lissy
30857cc288 Switch training model dependency to upstream-based
Currently, self-built Python package is not able to work, likely because
of the visibility of symbols. Let's switch to upstream TensorFlow for
now.
2018-01-31 16:54:37 +01:00
Alexandre Lissy
ade66b015d Build TensorFlow as monolithic into libdeepspeech.so
Also limit the set of included Ops to avoid wasting size.
2018-01-31 16:53:57 +01:00
Alexandre Lissy
b633048dff Making v0.1.1 release 2018-01-31 13:18:41 +01:00
Alexandre Lissy
fee3e5cdf5 Switch to swig 3.0.10-1.2
Fixes #1183
2018-01-19 11:25:53 +01:00
Alexandre Lissy
b4591cd6f5 Switch URLs to commit-based
sed \
  -ri 's/pip\.master\./pip.master.1390dc180e25b5821be80b407ddc5fad73d4ef6a./g' \
  taskcluster/*.yml

Fixes #1177
2018-01-17 16:49:38 +01:00
Alexandre Lissy
968c22c392 Avoid over-linking deepspeech.utils Python wrapper
Fixes #1116
2017-12-19 20:33:10 +01:00
Alexandre Lissy
5fdd77d24e Add OSX testing
Fixes #1072
2017-12-07 15:48:11 +01:00
Alexandre Lissy
e0a96ef751 Set minimum OSX version to 10.10
Fixes #1055
2017-12-05 09:21:10 +01:00
Alexandre Lissy
fb456da394 Test training with multiple Python versions
Fixes #970
2017-12-01 22:43:34 +01:00
Alexandre Lissy
1c5aa0a881 Run training test against forked and upstream TensorFlow 2017-11-30 23:58:27 +01:00
Alexandre Lissy
5b0ca1e410 Update to TensorFlow r1.4
Fixes #959
2017-11-30 18:51:13 +01:00
Alexandre Lissy
79d381f1b9 Patch SWIG locally with PR #968
Fixes #911
2017-11-30 16:56:25 +01:00