spaCy/spacy
Matthew Honnibal 2c37e0ccf6
💫 Use Blis for matrix multiplications (#2966)
Our epic matrix multiplication odyssey is drawing to a close...

I've now finally got the Blis linear algebra routines in a self-contained Python package, with wheels for Windows, Linux and OSX. The only missing platform at the moment is Windows Python 2.7. The result is at https://github.com/explosion/cython-blis

Thinc v7.0.0 will make the change to Blis. I've put a Thinc v7.0.0.dev0 up on PyPi so that we can test these changes with the CI, and even get them out to spacy-nightly, before Thinc v7.0.0 is released. This PR also updates the other dependencies to be in line with the current versions master is using. I've also resolved the msgpack deprecation problems, and gotten spaCy and Thinc up to date with the latest Cython.

The point of switching to Blis is to have control of how our matrix multiplications are executed across platforms. When we were using numpy for this, a different library would be used on pip and conda, OSX would use Accelerate, etc. This would open up different bugs and performance problems, especially when multi-threading was introduced.

With the change to Blis, we now strictly single-thread the matrix multiplications. This will make it much easier to use multiprocessing to parallelise the runtime, since we won't have nested parallelism problems to deal with.

* Use blis

* Use -2 arg to Cython

* Update dependencies

* Fix requirements

* Update setup dependencies

* Fix requirement typo

* Fix msgpack errors

* Remove Python27 test from Appveyor, until Blis works there

* Auto-format setup.py

* Fix murmurhash version
2018-11-27 00:44:04 +01:00
..
cli Merge branch 'develop' of https://github.com/explosion/spaCy into develop 2018-11-16 13:08:22 +01:00
data
displacy fix issue #2452 - displacy arrow direction is always forward (#2506) (closes #2452) 2018-07-04 14:12:08 +02:00
lang 💫 Make like_num work for prefixed numbers (#2808) 2018-10-01 10:49:14 +02:00
syntax 💫 Use Blis for matrix multiplications (#2966) 2018-11-27 00:44:04 +01:00
tests Tidy up [ci skip] 2018-11-26 18:56:04 +01:00
tokens Fix copy_reg compatibility on _serialize module 2018-09-28 15:23:14 +02:00
__init__.pxd
__init__.py Try again to filter warnings 2018-08-10 00:42:54 +02:00
__main__.py 💫 Add experimental ULMFit/BERT/Elmo-like pretraining (#2931) 2018-11-15 22:17:16 +01:00
_align.pyx Improve alignment around quotes 2018-08-16 01:04:34 +02:00
_ml.py Fix tensorizer 2018-11-02 23:29:54 +00:00
about.py Set version to 2.1.0a2 2018-11-08 09:56:56 +01:00
attrs.pxd Fix LANG symbol 2018-02-17 18:10:50 +01:00
attrs.pyx Revert "Merge branch 'develop' of https://github.com/explosion/spaCy into develop" 2018-03-27 19:23:02 +02:00
compat.py Port over #2949 2018-11-26 18:54:27 +01:00
errors.py Add deprecation warning for PhraseMatcher max_length 2018-11-14 19:10:46 +01:00
glossary.py Fix typo in glossary (resolves #1964) 2018-02-10 11:58:41 +01:00
gold.pxd
gold.pyx Lower-case text before alignment 2018-08-16 00:42:36 +02:00
language.py Small hack in Language.update to make torch work 2018-09-13 22:51:52 +00:00
lemmatizer.py Fix lemmatization 2018-07-05 13:56:02 +02:00
lexeme.pxd
lexeme.pyx 💫 Add .similarity warnings for no vectors and option to exclude warnings (#2197) 2018-05-21 01:22:38 +02:00
matcher.pyx 💫 Allow matching non-ORTH attributes in PhraseMatcher (#2925) 2018-11-15 03:00:58 +01:00
morphology.pxd Revert "Merge branch 'develop' of https://github.com/explosion/spaCy into develop" 2018-03-27 19:23:02 +02:00
morphology.pyx Fix lemmatization 2018-07-05 13:56:02 +02:00
parts_of_speech.pxd
parts_of_speech.pyx Tidy up rest 2017-10-27 21:07:59 +02:00
pipeline.pxd Fix names of pipeline components 2017-10-26 12:38:23 +02:00
pipeline.pyx 💫 Use Blis for matrix multiplications (#2966) 2018-11-27 00:44:04 +01:00
scorer.py Fix scoring if tokenization changes 2018-05-01 01:33:20 +02:00
strings.pxd Try to fix StringStore clean up (see #1506) 2017-11-11 03:11:27 +03:00
strings.pyx 💫 New system for error messages and warnings (#2163) 2018-04-03 15:50:31 +02:00
structs.pxd Make TokenC.sent_tart an int, to allow ternary value 2017-10-08 19:58:54 +02:00
symbols.pxd Revert "Merge branch 'develop' of https://github.com/explosion/spaCy into develop" 2018-03-27 19:23:02 +02:00
symbols.pyx Revert "Merge branch 'develop' of https://github.com/explosion/spaCy into develop" 2018-03-27 19:23:02 +02:00
tokenizer.pxd Disable tokenizer cache for special-cases. Fixes #1250 2017-10-24 16:08:05 +02:00
tokenizer.pyx 💫 Small efficiency fixes to tokenizer (#2587) 2018-07-24 23:35:54 +02:00
typedefs.pxd
typedefs.pyx Tidy up rest 2017-10-27 21:07:59 +02:00
util.py 💫 Use Blis for matrix multiplications (#2966) 2018-11-27 00:44:04 +01:00
vectors.pyx 💫 New system for error messages and warnings (#2163) 2018-04-03 15:50:31 +02:00
vocab.pxd 💫 Small efficiency fixes to tokenizer (#2587) 2018-07-24 23:35:54 +02:00
vocab.pyx Fix error in vocab 2018-08-16 17:18:09 +02:00