Paul O'Leary McCann
196886bbca
Fix coref size inference ( #10916 )
...
* Add explicit tok2vec_size parameter in clusterer
* Add tok2vec size to span predictor config
* Minor fixes
2022-06-08 20:03:41 +09:00
svlandeg
cea40c9d7b
fix types + black formatting
2022-05-25 13:34:09 +02:00
Adriane Boyd
f75a528787
Update spacy/ml/models/spancat.py
2022-05-25 13:05:41 +02:00
svlandeg
015050f42c
Merge branch 'master' into feature/coref
2022-05-25 13:01:56 +02:00
Paul O'Leary McCann
838f50192b
Black formatting
2022-05-25 19:20:03 +09:00
Paul O'Leary McCann
2a8efda689
Code review suggestions, cleanup
2022-05-25 19:18:26 +09:00
Paul O'Leary McCann
e721c7bed8
Import cleanup
2022-05-25 19:12:20 +09:00
Richard Hudson
32954c3bcb
Fix issues for Mypy 0.950 and Pydantic 1.9.0 ( #10786 )
...
* Make changes to typing
* Correction
* Format with black
* Corrections based on review
* Bumped Thinc dependency version
* Bumped blis requirement
* Correction for older Python versions
* Update spacy/ml/models/textcat.py
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
* Corrections based on review feedback
* Readd deleted docstring line
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
2022-05-25 09:33:54 +02:00
Paul O'Leary McCann
c9233a5a1f
Import torch from thinc
2022-05-24 17:28:27 +09:00
Paul O'Leary McCann
5cbc9f4573
Use thinc.util.has_torch
2022-05-24 16:02:39 +09:00
Paul O'Leary McCann
b1118cee58
Move epsilon
2022-05-24 15:59:08 +09:00
Paul O'Leary McCann
9da16df96e
Add guards around torch import
...
Torch is required for the coref/spanpred models but shouldn't be
required for spaCy in general.
The one tricky part of this is that one function in coref_util relied on
torch, but that file was imported in several places. Since the function
was only used in one place I moved it there.
2022-05-24 15:16:25 +09:00
kadarakos
1dc3894447
new parameters
2022-05-17 15:36:32 +00:00
kadarakos
403fb95d56
merge
2022-05-17 06:56:34 +00:00
Paul O'Leary McCann
2e8f0e9168
Rename coref params
2022-05-16 16:50:10 +09:00
Paul O'Leary McCann
13481fbcc2
Remove unused param, add TODOs about typing
2022-05-13 19:29:28 +09:00
kadarakos
b7ac4b33e2
fixing arguments
2022-05-11 14:59:59 +00:00
kadarakos
7cf6bcca0e
merge misery
2022-05-10 17:19:16 +00:00
kadarakos
e512874c80
small refactor and docs
2022-05-10 16:40:31 +00:00
Paul O'Leary McCann
33f4f90ff0
Formatting
2022-05-10 19:09:52 +09:00
Paul O'Leary McCann
41fc092674
Split span predictor model into its own file
2022-05-10 19:08:21 +09:00
svlandeg
6b51258a58
clean up unused imports + black formatting
2022-05-09 13:34:50 +02:00
Raphael Mitsch
f5390e278a
Refactor error messages to remove hardcoded strings ( #10729 )
...
* Use custom error msg instead of hardcoded string: replaced remaining hardcoded error message strings.
* Use custom error msg instead of hardcoded string: fixing faulty Errors import.
2022-05-02 13:38:46 +02:00
Paul O'Leary McCann
683f470852
Merge branch 'master' into feature/coref
2022-04-18 18:39:08 +09:00
kadarakos
b53113e3b8
Preparing span predictor for predicting from gold ( #10547 )
...
Note this is squashed because rebasing had conflicts.
* remove unnecessary .device
* span predictor debug start
* gearing up SpanPredictor for gold-heads
* merge SpanPredictor attributes
* remove useless extra prefix and device from spanpredictor
* make sure predicted and reference keeps aligned
* handle empty head_ids
* handle empty clusters
* addressing suggestions by @polm
* nicer restore
* fix score overwriting bug
* prepare for aligned heads-spans training
* span accuracy score
* update with eg.predited as other components
* add backprop callback to spanpredictor
* report start- and end-accuracies separately
* fixing scorer
Co-authored-by: Kádár Ákos <akos@onyx.uvt.nl>
2022-04-13 19:42:49 +09:00
Kádár Ákos
2a1ad4c5d2
add backprop callback to spanpredictor
2022-04-08 14:56:44 +02:00
Kádár Ákos
4fc40340f9
handle empty head_ids
2022-03-28 11:28:21 +02:00
Kádár Ákos
83ac0477c8
remove useless extra prefix and device from spanpredictor
2022-03-24 16:44:50 +01:00
Kádár Ákos
1c5dabcb47
merge SpanPredictor attributes
2022-03-24 16:23:12 +01:00
Kádár Ákos
a872c69ffb
merge
2022-03-24 16:10:04 +01:00
Kádár Ákos
706b2e6f25
gearing up SpanPredictor for gold-heads
2022-03-24 16:06:20 +01:00
Kádár Ákos
150e7c46d7
conflict
2022-03-23 11:27:02 +01:00
Kádár Ákos
1eaf8fb0cf
span predictor debug start
2022-03-23 11:24:27 +01:00
Paul O'Leary McCann
eec00ce60d
Fix various sizes in SpanPredictor FFNN
2022-03-23 16:20:31 +09:00
Paul O'Leary McCann
2190cbc0e6
Add progress on SpanPredictor component
...
This isn't working. There is a CUDA error in the torch code during
initialization and it's not clear why.
2022-03-19 19:39:49 +09:00
Kádár Ákos
db422abf01
remove unnecessary .device
2022-03-18 16:24:26 +01:00
Paul O'Leary McCann
0275ae29de
Remove stale comment
2022-03-16 20:09:12 +09:00
Paul O'Leary McCann
6974f55daa
Hack for transformer listener size
2022-03-16 15:15:53 +09:00
Paul O'Leary McCann
5650853c0f
Remove unused functions
2022-03-16 14:38:11 +09:00
Daniël de Kok
e5debc68e4
Tagger: use unnormalized probabilities for inference ( #10197 )
...
* Tagger: use unnormalized probabilities for inference
Using unnormalized softmax avoids use of the relatively expensive exp function,
which can significantly speed up non-transformer models (e.g. I got a speedup
of 27% on a German tagging + parsing pipeline).
* Add spacy.Tagger.v2 with configurable normalization
Normalization of probabilities is disabled by default to improve
performance.
* Update documentation, models, and tests to spacy.Tagger.v2
* Move Tagger.v1 to spacy-legacy
* docs/architectures: run prettier
* Unnormalized softmax is now a Softmax_v2 option
* Require thinc 8.0.14 and spacy-legacy 3.0.9
2022-03-15 14:15:31 +01:00
Paul O'Leary McCann
d0ae2590db
Delete all the coref-hoi code
2022-03-15 20:05:24 +09:00
Paul O'Leary McCann
abdc7d87af
Clean up util code
...
Moved everything into coref_util.py, deleted wl-specific file.
2022-03-15 19:59:44 +09:00
Paul O'Leary McCann
0522a43116
Make span2head component
2022-03-15 19:19:15 +09:00
Paul O'Leary McCann
e6917d8dc4
Add util functions for wl-coref
2022-03-14 19:27:55 +09:00
Paul O'Leary McCann
8eadf3781b
Training runs now
...
Evaluation needs fixing, and code still needs cleanup.
2022-03-14 19:02:17 +09:00
Paul O'Leary McCann
d22a002641
Forward/backward pass works
...
Evaluate does not work - predict hasn't been updated
2022-03-14 17:26:27 +09:00
Paul O'Leary McCann
c4f9c24738
The coref model is able to be loaded
...
The span predictor component is initialized but not used at all now.
Plan is to work on it after the word level clustering part is trainable
end-to-end.
2022-03-09 19:31:11 +09:00
Paul O'Leary McCann
35cc2b138f
Add span predictor code
...
Accidentally omitted before
2022-03-08 18:13:26 +09:00
Paul O'Leary McCann
1c697b4011
Remove references to config
...
Replaced with model arguments
2022-03-08 18:13:09 +09:00
Paul O'Leary McCann
c0cd5025e3
Start bringin in wl-coref
...
This absolutely does not work. First step here is getting over most of
the code in roughly the files we want it in. After the code has been
pulled over it can be restructured to match spaCy and cleaned up.
2022-03-06 20:00:15 +09:00
