honnibal Follow

Matthew Honnibal honnibal

Computational linguist from Sydney and Berlin. Author of the spaCy NLP library.

1.5k followers · 5 following · 53

Highlights

Arctic Code Vault Contributor

Organizations

Popular repositories

spacy-pretrain-polyaxon

Example using Polyaxon to experiment with pre-training spaCy

Python 63 9
text_classification

Relatively simple text classification powered by spaCy

Python 40 16
dsr16_nlp

NLP tutorial for the Berlin Data Science Retreat

Python 40 9
stanford_corenlp_pywrapper

Forked from brendano/stanford_corenlp_pywrapper

Java 10 3
py-clearnlp-converter

A simple Python wrapper for the ClearNLP constituents-to-dependencies converter

Python 10 5
cython-blis

Cython wrapper for the BLIS linear algebra routines. Goal: fast BLAS off PyPi, no system dependency

C 8 3

1,390 contributions in the last year

Contribution activity

September 2020

Created a pull request in explosion/spaCy that received 4 comments

Sep 8

Improve v3 pretrain command

We intentionally let the spacy pretrain command fall behind our conventions a bit in v3, while we let the rest of the config system stabilise. This…

+158 −187 • 4 comments

Support data augmentation in Corpus Sep 27
Make a pre-check to speed up alignment cache Sep 24
Fix skipped documents in entity scorer Sep 24
Make project pull order insensitive Sep 23
Improve subprocess handling and git clone Sep 20
Temporary work-around for scoring a subset of components Sep 18
Draft docs for pretrain usage Sep 17
Draft text for static vectors intro Sep 17
Fix conversion of older CoNLL parsing files Sep 12
Fix tagger training when some tags are missing Sep 12
Add quick hack to allow CLI extensions Sep 8
Fix significant performance bug in parser training Sep 1
Remove 'cleanup' of strings Sep 1

Add Model methods to check deserialization compatibility Sep 28
Remove jax for now Sep 24

Update to new spacy-nightly Sep 8

Support data augmentation in Corpus Sep 27
Fix skipped documents in entity scorer Sep 24
Modify setting missing and blocked entity tokens Sep 24
Fix handling of score_weights Sep 24
Add option to disable Matcher errors Sep 24
extend quickstart components Sep 22
Clean up spacy.tokens Sep 16
More extensible corpus in config Sep 16
Add vectors option to CharacterEmbed Sep 15
fix W&B logger Sep 15
Handle config and meta correctly in spacy.blank Sep 15
Fix sparse checkout and error handling Sep 14

fix model.from_dict Sep 24

Created an issue in wandb/client that received 2 comments

Sep 5

Allow setting executable via environment variable

Hey :) In _init_headless there are some comments about using sys.executable to execute the subprocess: https://github.com/wandb/client/blob/master/…

2 comments

Preliminary discussion: Standardise support for reuse of buffers Sep 21