Skip to content

Releases: piskvorky/gensim

4.3.2

24 Aug 08:44
@mpenkov mpenkov
4.3.2
30179c8
Compare
Choose a tag to compare
4.3.2 Latest
Latest

Changes

4.3.2, 2023-08-23

🔴 Bug fixes

  • Fix incorrect conversion of cosine distance to cosine similarity (monash849, #3441)

📚 Tutorial and doc improvements

👍 Improvements

🔮 Testing, CI, housekeeping

Assets 2
Loading

4.3.0

21 Dec 00:42
@mpenkov mpenkov
4.3.0
adf393c
Compare
Choose a tag to compare
4.3.0

What's Changed

  • Allow overriding the Cython version requirement by @pabs3 in #3323
  • Update Python module MANIFEST by @pabs3 in #3343
  • Clean up references to Morfessor, tox and gensim.models.wrappers by @pabs3 in #3345
  • Disable the Gensim 3=>4 warning in docs by @piskvorky in #3346
  • pin sphinx versions, add explicit gallery_top label by @mpenkov in #3383
  • Declare variables prior to for loop in fastss.pyx for ANSI C compatibility by @hstk30 in #3378
  • Fix typo in word2vec and KeyedVectors docstrings by @dymil in #3365
  • Replace np.multiply with np.square and copyedit in translation_matrix.py by @dymil in #3374
  • Copyedit and fix outdated statements in translation matrix tutorial by @dymil in #3375
  • Implement Okapi BM25 variants in Gensim by @Witiko in #3304
  • Giving missing credit in EnsembleLDA to Alex in docs by @sezanzeb in #3393
  • PERF: pyemd to POT for EMD computation in wmdistance by @TLouf in #3327
  • Fixed bug in loss computation for Word2Vec with hierarchical softmax by @TalIfargan in #3397
  • fix deprecation warning from pytest by @martino-vic in #3354
  • Switch to Cython language level 3 by @pabs3 in #3344
  • Implement numpy hack in setup.py to enable install under Poetry by @jaymegordo in #3363
  • Fixed the broken link in readme.md by @aswin2108 in #3409
  • Path Coherence Model to correctly handle empty documents by @PrimozGodec in #3406
  • Add support for Python 3.11 and drop support for Python 3.7 by @acul3 in #3402
  • clarify runtime expectations by @gojomo in #3381
  • Fix bug that prevents loading old models by @funasshi in #3359
  • refactor wheel building and testing workflow by @mpenkov in #3410
  • Fixed FastTextKeyedVectors handling in add_vector by @globba in #3389
  • Flsamodel by @ERijck in #3398
  • Fix backwards compatibility bug in Word2Vec by @mpenkov in #3415
  • fix numpy hack in setup.py by @mpenkov in #3416
  • updated changelog for next release by @mpenkov in #3412

New Contributors

Full Changelog: 4.2.0...4.3.0

Contributors

gojomo, pabs3, and 16 other contributors
Assets 2
Loading

4.2.0

01 May 08:38
@piskvorky piskvorky
4.2.0
acbba2f
Compare
Choose a tag to compare
4.2.0

A number of incremental improvements, optimizations and bugfixes: CHANGELOG

Assets 2
Loading

4.1.2

18 Sep 14:21
@mpenkov mpenkov
4.1.2
b76108e
Compare
Choose a tag to compare
4.1.2

4.1.2, 2021-09-17

This is a bugfix release that addresses left over compatibility issues with older versions of numpy and MacOS.

4.1.1, 2021-09-14

This is a bugfix release that addresses compatibility issues with older versions of numpy.

4.1.0, 2021-08-15

Gensim 4.1 brings two major new functionalities:

There are several minor changes that are not backwards compatible with previous versions of Gensim.
The affected functionality is relatively less used, so it is unlikely to affect most users, so we have opted to not require a major version bump.
Nevertheless, we describe them below.

Improved parameter edge-case handling in KeyedVectors most_similar and most_similar_cosmul methods

We now handle both positive and negative keyword parameters consistently.
They may now be either:

  1. A string, in which case the value is reinterpreted as a list of one element (the string value)
  2. A vector, in which case the value is reinterpreted as a list of one element (the vector)
  3. A list of strings
  4. A list of vectors

So you can now simply do:

    model.most_similar(positive='war', negative='peace')

instead of the slightly more involved

model.most_similar(positive=['war'], negative=['peace'])

Both invocations remain correct, so you can use whichever is most convenient.
If you were somehow expecting gensim to interpret the strings as a list of characters, e.g.

model.most_similar(positive=['w', 'a', 'r'], negative=['p', 'e', 'a', 'c', 'e'])

then you will need to specify the lists explicitly in gensim 4.1.

Deprecated obsolete step parameter from doc2vec

With the newer version, do this:

model.infer_vector(..., epochs=123)

instead of this:

model.infer_vector(..., steps=123)

Plus a large number of smaller improvements and fixes, as usual.

⚠️ If migrating from old Gensim 3.x, read the Migration guide first.

👍 New features

  • #3169: Implement shrink_windows argument for Word2Vec, by @M-Demay
  • #3163: Optimize word mover distance (WMD) computation, by @flowlight0
  • #3157: New KeyedVectors.vectors_for_all method for vectorizing all words in a dictionary, by @Witiko
  • #3153: Vectorize word2vec.predict_output_word for speed, by @M-Demay
  • #3146: Use FastSS for fast kNN over Levenshtein distance, by @Witiko
  • #3128: Materialize and copy the corpus passed to SoftCosineSimilarity, by @Witiko
  • #3115: Make LSI dispatcher CLI param for number of jobs optional, by @robguinness
  • #3091: LsiModel: Only log top words that actually exist in the dictionary, by @kmurphy4
  • #2980: Added EnsembleLda for stable LDA topics, by @sezanzeb
  • #2978: Optimize performance of Author-Topic model, by @horpto
  • #3000: Tidy up KeyedVectors.most_similar() API, by @simonwiles

📚 Tutorials and docs

🔴 Bug fixes

  • #3178: Fix Unicode string incompatibility in gensim.similarities.fastss.editdist, by @Witiko
  • #3174: Fix loading Phraser models stored in Gensim 3.x into Gensim 4.0, by @emgucv
  • #3136: Fix indexing error in word2vec_inner.pyx, by @bluekura
  • #3131: Add missing import to NMF docs and models/init.py, by @properGrammar
  • #3116: Fix bug where saved Phrases model did not load its connector_words, by @aloknayak29
  • #2830: Fixed KeyError in coherence model, by @pietrotrope

⚠️ Removed functionality & deprecations

  • #3176: Eliminate obsolete step parameter from doc2vec infer_vector and similarity_unseen_docs, by @rock420
  • #2965: Remove strip_punctuation2 alias of strip_punctuation, by @sciatro
  • #3180: Move preprocessing functions from gensim.corpora.textcorpus and gensim.corpora.lowcorpus to gensim.parsing.preprocessing, by @rock420

🔮 Testing, CI, housekeeping

  • #3156: Update Numpy minimum version to 1.17.0, by @PrimozGodec
  • #3143: replace _mul function with explicit casts, by @mpenkov
  • #2952: Allow newer versions of the Morfessor module for the tests, by @pabs3
  • #2965: Remove strip_punctuation2 alias of strip_punctuation, by @sciatro
Assets 2
Loading

4.1.1

14 Sep 13:49
@mpenkov mpenkov
4.1.1
3d72896
Compare
Choose a tag to compare
4.1.1

4.1.1, 2021-09-14

This is a bugfix release that addresses compatibility issues with older versions of numpy.

4.1.0, 2021-08-15

Gensim 4.1 brings two major new functionalities:

There are several minor changes that are not backwards compatible with previous versions of Gensim.
The affected functionality is relatively less used, so it is unlikely to affect most users, so we have opted to not require a major version bump.
Nevertheless, we describe them below.

Improved parameter edge-case handling in KeyedVectors most_similar and most_similar_cosmul methods

We now handle both positive and negative keyword parameters consistently.
They may now be either:

  1. A string, in which case the value is reinterpreted as a list of one element (the string value)
  2. A vector, in which case the value is reinterpreted as a list of one element (the vector)
  3. A list of strings
  4. A list of vectors

So you can now simply do:

    model.most_similar(positive='war', negative='peace')

instead of the slightly more involved

model.most_similar(positive=['war'], negative=['peace'])

Both invocations remain correct, so you can use whichever is most convenient.
If you were somehow expecting gensim to interpret the strings as a list of characters, e.g.

model.most_similar(positive=['w', 'a', 'r'], negative=['p', 'e', 'a', 'c', 'e'])

then you will need to specify the lists explicitly in gensim 4.1.

Deprecated obsolete step parameter from doc2vec

With the newer version, do this:

model.infer_vector(..., epochs=123)

instead of this:

model.infer_vector(..., steps=123)

Plus a large number of smaller improvements and fixes, as usual.

⚠️ If migrating from old Gensim 3.x, read the Migration guide first.

👍 New features

  • #3169: Implement shrink_windows argument for Word2Vec, by @M-Demay
  • #3163: Optimize word mover distance (WMD) computation, by @flowlight0
  • #3157: New KeyedVectors.vectors_for_all method for vectorizing all words in a dictionary, by @Witiko
  • #3153: Vectorize word2vec.predict_output_word for speed, by @M-Demay
  • #3146: Use FastSS for fast kNN over Levenshtein distance, by @Witiko
  • #3128: Materialize and copy the corpus passed to SoftCosineSimilarity, by @Witiko
  • #3115: Make LSI dispatcher CLI param for number of jobs optional, by @robguinness
  • #3091: LsiModel: Only log top words that actually exist in the dictionary, by @kmurphy4
  • #2980: Added EnsembleLda for stable LDA topics, by @sezanzeb
  • #2978: Optimize performance of Author-Topic model, by @horpto
  • #3000: Tidy up KeyedVectors.most_similar() API, by @simonwiles

📚 Tutorials and docs

🔴 Bug fixes

  • #3178: Fix Unicode string incompatibility in gensim.similarities.fastss.editdist, by @Witiko
  • #3174: Fix loading Phraser models stored in Gensim 3.x into Gensim 4.0, by @emgucv
  • #3136: Fix indexing error in word2vec_inner.pyx, by @bluekura
  • #3131: Add missing import to NMF docs and models/init.py, by @properGrammar
  • #3116: Fix bug where saved Phrases model did not load its connector_words, by @aloknayak29
  • #2830: Fixed KeyError in coherence model, by @pietrotrope

⚠️ Removed functionality & deprecations

  • #3176: Eliminate obsolete step parameter from doc2vec infer_vector and similarity_unseen_docs, by @rock420
  • #2965: Remove strip_punctuation2 alias of strip_punctuation, by @sciatro
  • #3180: Move preprocessing functions from gensim.corpora.textcorpus and gensim.corpora.lowcorpus to gensim.parsing.preprocessing, by @rock420

🔮 Testing, CI, housekeeping

  • #3156: Update Numpy minimum version to 1.17.0, by @PrimozGodec
  • #3143: replace _mul function with explicit casts, by @mpenkov
  • #2952: Allow newer versions of the Morfessor module for the tests, by @pabs3
  • #2965: Remove strip_punctuation2 alias of strip_punctuation, by @sciatro
Assets 2
Loading

4.1.0

29 Aug 22:28
@piskvorky piskvorky
4.1.0
109c88e
Compare
Choose a tag to compare
4.1.0

4.1.0, 2021-08-15

Gensim 4.1 brings two major new functionalities:

There are several minor changes that are not backwards compatible with previous versions of Gensim.
The affected functionality is relatively less used, so it is unlikely to affect most users, so we have opted to not require a major version bump.
Nevertheless, we describe them below.

Improved parameter edge-case handling in KeyedVectors most_similar and most_similar_cosmul methods

We now handle both positive and negative keyword parameters consistently.
They may now be either:

  1. A string, in which case the value is reinterpreted as a list of one element (the string value)
  2. A vector, in which case the value is reinterpreted as a list of one element (the vector)
  3. A list of strings
  4. A list of vectors

So you can now simply do:

    model.most_similar(positive='war', negative='peace')

instead of the slightly more involved

model.most_similar(positive=['war'], negative=['peace'])

Both invocations remain correct, so you can use whichever is most convenient.
If you were somehow expecting gensim to interpret the strings as a list of characters, e.g.

model.most_similar(positive=['w', 'a', 'r'], negative=['p', 'e', 'a', 'c', 'e'])

then you will need to specify the lists explicitly in gensim 4.1.

Deprecated obsolete step parameter from doc2vec

With the newer version, do this:

model.infer_vector(..., epochs=123)

instead of this:

model.infer_vector(..., steps=123)

Plus a large number of smaller improvements and fixes, as usual.

⚠️ If migrating from old Gensim 3.x, read the Migration guide first.

👍 New features

  • #3169: Implement shrink_windows argument for Word2Vec, by @M-Demay
  • #3163: Optimize word mover distance (WMD) computation, by @flowlight0
  • #3157: New KeyedVectors.vectors_for_all method for vectorizing all words in a dictionary, by @Witiko
  • #3153: Vectorize word2vec.predict_output_word for speed, by @M-Demay
  • #3146: Use FastSS for fast kNN over Levenshtein distance, by @Witiko
  • #3128: Materialize and copy the corpus passed to SoftCosineSimilarity, by @Witiko
  • #3115: Make LSI dispatcher CLI param for number of jobs optional, by @robguinness
  • #3091: LsiModel: Only log top words that actually exist in the dictionary, by @kmurphy4
  • #2980: Added EnsembleLda for stable LDA topics, by @sezanzeb
  • #2978: Optimize performance of Author-Topic model, by @horpto
  • #3000: Tidy up KeyedVectors.most_similar() API, by @simonwiles

📚 Tutorials and docs

🔴 Bug fixes

  • #3178: Fix Unicode string incompatibility in gensim.similarities.fastss.editdist, by @Witiko
  • #3174: Fix loading Phraser models stored in Gensim 3.x into Gensim 4.0, by @emgucv
  • #3136: Fix indexing error in word2vec_inner.pyx, by @bluekura
  • #3131: Add missing import to NMF docs and models/init.py, by @properGrammar
  • #3116: Fix bug where saved Phrases model did not load its connector_words, by @aloknayak29
  • #2830: Fixed KeyError in coherence model, by @pietrotrope

⚠️ Removed functionality & deprecations

  • #3176: Eliminate obsolete step parameter from doc2vec infer_vector and similarity_unseen_docs, by @rock420
  • #2965: Remove strip_punctuation2 alias of strip_punctuation, by @sciatro
  • #3180: Move preprocessing functions from gensim.corpora.textcorpus and gensim.corpora.lowcorpus to gensim.parsing.preprocessing, by @rock420

🔮 Testing, CI, housekeeping

  • #3156: Update Numpy minimum version to 1.17.0, by @PrimozGodec
  • #3143: replace _mul function with explicit casts, by @mpenkov
  • #2952: Allow newer versions of the Morfessor module for the tests, by @pabs3
  • #2965: Remove strip_punctuation2 alias of strip_punctuation, by @sciatro

4.0.1, 2021-04-01

Bugfix release to address issues with Wheels on Windows:

4.0.0, 2021-03-24

⚠️ Gensim 4.0 contains breaking API changes! See the Migration guide to update your existing Gensim 3.x code and models.

Gensim 4.0 is a major release with lots of performance & robustness improvements, and a new website.

Main highlights

  • Massively optimized popular algorithms the community has grown to love: fastText, word2vec, doc2vec, phrases:

    a. Efficiency

    | model | 3.8.3: wall time / peak RAM / throughput ...

Read more
Assets 2
Loading

4.0.1

01 Apr 14:19
@mpenkov mpenkov
4.0.1
b4f64a9
Compare
Choose a tag to compare
4.0.1

4.0.1, 2021-04-01

Bugfix release to address issues with wheels on Windows due to Numpy binary incompatibility:

4.0.0, 2021-03-24

⚠️ Gensim 4.0 contains breaking API changes! See the Migration guide to update your existing Gensim 3.x code and models.

Gensim 4.0 is a major release with lots of performance & robustness improvements, and a new website.

Main highlights

  • Massively optimized popular algorithms the community has grown to love: fastText, word2vec, doc2vec, phrases:

    a. Efficiency

    model 3.8.3: wall time / peak RAM / throughput 4.0.0: wall time / peak RAM / throughput
    fastText 2.9h / 4.11 GB / 822k words/s 2.3h / 1.26 GB / 914k words/s
    word2vec 1.7h / 0.36 GB / 1685k words/s 1.2h / 0.33 GB / 1762k words/s

    In other words, fastText now needs 3x less RAM (and is faster); word2vec has 2x faster init (and needs less RAM, and is faster); detecting collocation phrases is 2x faster. (4.0 benchmarks)

    b. Robustness. We fixed a bunch of long-standing bugs by refactoring the internal code structure (see 🔴 Bug fixes below)

    c. Simplified OOP model for easier model exports and integration with TensorFlow, PyTorch &co.

    These improvements come to you transparently aka "for free", but see Migration guide for some changes that break the old Gensim 3.x API. Update your code accordingly.

  • Dropped a bunch of externally contributed modules and wrappers: summarization, pivoted TFIDF, Mallet…

    • Code quality was not up to our standards. Also there was no one to maintain these modules, answer user questions, support them.

      So rather than let them rot, we took the hard decision of removing these contributed modules from Gensim. If anyone's interested in maintaining them, please fork & publish into your own repo. They can live happily outside of Gensim.

  • Dropped Python 2. Gensim 4.0 is Py3.6+. Read our Python version support policy.

    • If you still need Python 2 for some reason, stay at Gensim 3.8.3.

  • A new Gensim website – finally! 🙃

So, a major clean-up release overall. We're happy with this tighter, leaner and faster Gensim.

This is the direction we'll keep going forward: less kitchen-sink of "latest academic algorithms", more focus on robust engineering, targetting concrete NLP & document similarity use-cases.

👍 New features

📚 Tutorials and docs

🔴 Bug fixes

Read more
Assets 2
Loading

4.0.0

25 Mar 13:42
@mpenkov mpenkov
4.0.0
f46d72a
Compare
Choose a tag to compare
4.0.0

Changes

4.0.0, 2021-03-24

⚠️ Gensim 4.0 contains breaking API changes! See the Migration guide to update your existing Gensim 3.x code and models.

Gensim 4.0 is a major release with lots of performance & robustness improvements, and a new website.

Main highlights

  • Massively optimized popular algorithms the community has grown to love: fastText, word2vec, doc2vec, phrases:

    a. Efficiency

    model 3.8.3: wall time / peak RAM / throughput 4.0.0: wall time / peak RAM / throughput
    fastText 2.9h / 4.11 GB / 822k words/s 2.3h / 1.26 GB / 914k words/s
    word2vec 1.7h / 0.36 GB / 1685k words/s 1.2h / 0.33 GB / 1762k words/s

    In other words, fastText now needs 3x less RAM (and is faster); word2vec has 2x faster init (and needs less RAM, and is faster); detecting collocation phrases is 2x faster. (4.0 benchmarks)

    b. Robustness. We fixed a bunch of long-standing bugs by refactoring the internal code structure (see 🔴 Bug fixes below)

    c. Simplified OOP model for easier model exports and integration with TensorFlow, PyTorch &co.

    These improvements come to you transparently aka "for free", but see Migration guide for some changes that break the old Gensim 3.x API. Update your code accordingly.

  • Dropped a bunch of externally contributed modules and wrappers: summarization, pivoted TFIDF, Mallet…

    • Code quality was not up to our standards. Also there was no one to maintain these modules, answer user questions, support them.

      So rather than let them rot, we took the hard decision of removing these contributed modules from Gensim. If anyone's interested in maintaining them, please fork & publish into your own repo. They can live happily outside of Gensim.

  • Dropped Python 2. Gensim 4.0 is Py3.6+. Read our Python version support policy.

    • If you still need Python 2 for some reason, stay at Gensim 3.8.3.

  • A new Gensim website – finally! 🙃

So, a major clean-up release overall. We're happy with this tighter, leaner and faster Gensim.

This is the direction we'll keep going forward: less kitchen-sink of "latest academic algorithms", more focus on robust engineering, targetting concrete NLP & document similarity use-cases.

👍 New features

📚 Tutorials and docs

🔴 Bug fixes

  • #2891: Fix fastText word-vectors with ngrams off, by @gojomo
  • #2907: Fix doc2vec crash for large sets of doc-vectors, by @gojomo
  • #2899: Fix similarity bug in NMSLIB indexer, by @piskvorky
  • #2899: Fix deprecation warnings in Annoy integration, by @piskvorky
  • #2901: Fix inheritance of WikiCorpus from TextCorpus, by @jenishah
  • #2940: Fix deprecations in SoftCosineSimilarity, by @Witiko
  • #2944: Fix save_facebook_model failure after update-vocab & other initialization streamlining, by @gojomo
  • #2846: Fix for Python 3.9/3.10: remove xml.etree.cElementTree, by @hugovk
  • #2973: phrases.export_phrases() doesn't yield all bigrams, by @piskvorky
  • #2942: Segfault when training doc2vec, by @gojomo
  • #3041: Fix RuntimeError in export_phrases (change defaultdict to dict), by @thalishsajeed
  • #3059: Fix rac...
Read more
Assets 2
Loading

4.0.0.rc1

22 Mar 09:16
@mpenkov mpenkov
4.0.0.rc1
4a241f0
Compare
Choose a tag to compare
4.0.0.rc1 Pre-release
Pre-release

4.0.0.rc1, 2021-03-19

⚠️ Gensim 4.0 contains breaking API changes! See the Migration guide to update your existing Gensim 3.x code and models.

Gensim 4.0 is a major release with lots of performance & robustness improvements and a new website.

Main highlights (see also 👍 Improvements below)

  • Massively optimized popular algorithms the community has grown to love: fastText, word2vec, doc2vec, phrases:

    a. Efficiency

    model 3.8.3: wall time / peak RAM / throughput 4.0.0: wall time / peak RAM / throughput
    fastText 2.9h / 4.11 GB / 822k words/s 2.3h / 1.26 GB / 914k words/s
    word2vec 1.7h / 0.36 GB / 1685k words/s 1.2h / 0.33 GB / 1762k words/s

    In other words, fastText now needs 3x less RAM (and is faster); word2vec has 2x faster init (and needs less RAM, and is faster); detecting collocation phrases is 2x faster. (4.0 benchmarks)

    b. Robustness. We fixed a bunch of long-standing bugs by refactoring the internal code structure (see 🔴 Bug fixes below)

    c. Simplified OOP model for easier model exports and integration with TensorFlow, PyTorch &co.

    These improvements come to you transparently aka "for free", but see Migration guide for some changes that break the old Gensim 3.x API. Update your code accordingly.

  • Dropped a bunch of externally contributed modules: summarization, pivoted TFIDF normalization, FIXME.

    • Code quality was not up to our standards. Also there was no one to maintain them, answer user questions, support these modules.

      So rather than let them rot, we took the hard decision of removing these contributed modules from Gensim. If anyone's interested in maintaining them please fork into your own repo, they can live happily outside of Gensim.

  • Dropped Python 2. Gensim 4.0 is Py3.6+. Read our Python version support policy.

    • If you still need Python 2 for some reason, stay at Gensim 3.8.3.

  • A new Gensim website – finally! 🙃

So, a major clean-up release overall. We're happy with this tighter, leaner and faster Gensim.

This is the direction we'll keep going forward: less kitchen-sink of "latest academic algorithms", more focus on robust engineering, targetting common concrete NLP & document similarity use-cases.

🌟 New Features

🔴 Bug fixes

📚 Tutorial and doc improvements

  • fix various documentation warnings (mpenkov, #3077)
  • Fix broken link in run_doc how-to (sezanzeb, #2991)
  • Point WordEmbeddingSimilarityIndex documentation to gensim.similarities (Witiko, #3003)
  • Make the link to the Gensim 3.8.3 documentation dynamic (Witiko, #2996)

⚠️ Removed functionality

🔮 Miscellaneous

Assets 2
Loading

4.0.0beta

01 Nov 13:33
@mpenkov mpenkov
4.0.0beta
8624aa2
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
4.0.0beta Pre-release
Pre-release

4.0.0beta, 2020-10-31

⚠️ Gensim 4.0 contains breaking API changes! See the Migration guide to update your existing Gensim 3.x code and models.

Main highlights

  • Massively optimized popular algorithms the community has grown to love: fastText, word2vec, doc2vec, phrases:

    a. Efficiency

    model 3.8.3
    wall time / peak RAM / throughput    		 4.0.0
    wall time / peak RAM / throughput
    fastText 2.9h / 4.11 GB / 822k words/s 2.3h / 1.26 GB / 914k words/s
    word2vec 1.7h / 0.36 GB / 1685k words/s 1.2h / 0.33 GB / 1762k words/s

    In other words, fastText now needs 3x less RAM (and is faster); word2vec has 2x faster init (and needs less RAM, and is faster); detecting collocation phrases is 2x faster. 4.0 benchmarks.

    b. Robustness. We fixed a bunch of long-standing bugs by refactoring the internal code structure (see 🔴 Bug fixes below)

    c. Simplified OOP model for easier model exports and integration with TensorFlow, PyTorch &co.

    These improvements come to you transparently aka "for free", but see Migration guide for some changes that break the old Gensim 3.x API. Update your code accordingly.

  • Dropped a bunch of externally contributed modules: summarization, pivoted TFIDF normalization, wrappers for 3rd party libraries: Mallet, scikit-learn, DTM model, Vowpal Wabbit, wordrank, varembed.

    • Why? Code quality was not up to our standards. Also there was no one to maintain them, answer user questions, support these modules and wrappers.

      So rather than let them rot, we took the hard decision of removing these contributed modules from Gensim. If anyone's interested in maintaining them please fork into your own repo, they can live happily outside of Gensim, linked to as "contributed" from Gensim docs.

  • Dropped Python 2. Gensim 4.0 is Py3.6+. Read our Python version support policy.

    • If you still need Python 2 for some reason, stay at Gensim 3.8.3.

  • A new Gensim website – finally! 🙃

So, a major clean-up release overall. We're happy with this tighter, leaner and faster Gensim.

This is the direction we'll keep going forward: less kitchen-sink of "latest academic fad", more focus on robust engineering, targetting common NLP & document similarity use-cases.

Why a pre-release?

This 4.0.0beta pre-release is for users who want the cutting edge performance and bug fixes. Plus users who want to help out, by testing and providing feedback: code, documentation, workflows… Please let us know on the mailing list!

Install the pre-release with:

pip install --pre --upgrade gensim

What will change between this pre-release and a "full" 4.0 release?

Check progress here.

👍 Improvements

📚 Tutorials and docs

🔴 Bug fixes

  • #2891: Fix fastText word-vectors with ngrams off, by @gojomo
  • #2907: Fix doc2vec crash for large sets of doc-vectors, by @gojomo
  • #2899: Fix similarity bug in NMSLIB indexer, by @piskvorky
  • #2899: Fix deprecation warnings in Annoy integration, by @piskvorky
  • #2901: Fix inheritance of WikiCorpus from TextCorpus, by @jenishah
  • #2940; Fix deprecations in SoftCosineSimilarity, by @Witiko
  • #2944: Fix save_facebook_model failure after update-vocab & other initialization streamlining, by @gojomo
  • #2846: Fix for Python 3.9/3.10: remove xml.etree.cElementTree, by @hugovk
  • #2973: phrases.export_phrases() doesn't yield all bigrams
  • #2942: Segfault when training doc2vec

⚠️ Removed functionality & deprecations

  • #6: No more binary wheels for x32 platforms, by menshikh-iv
  • #2899: Renamed overly broad similarities.index to the more appropriate similarities.annoy, by @piskvorky
  • #2958: Remove gensim.summarization subpackage, docs and test data, by @mpenkov
  • #2926: Rename num_words to topn in dtm_coherence, by @MeganStodel
  • #2937: Remove Keras dependency, by @piskvorky
  • Removed all code, methods, attributes and functions marked as deprecated in Gensim 3.8.3.
Assets 2
Loading