Rooba
/
spaCy
mirror of https://github.com/explosion/spaCy.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity
spaCy/spacy/cli/templates/quickstart_training_recomme...

122 lines
2.8 KiB
YAML
Raw Normal View History

Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
# Recommended settings and available resources for each language, if available.
fix typo
2020-08-19 16:46:08 +00:00
# Not all languages have recommended word vectors or transformers and for some,
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
# the recommended transformer for efficiency and accuracy may be the same.
en:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: en_core_web_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
name: roberta-base
size_factor: 3
accuracy:
name: roberta-base
size_factor: 3
de:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: de_core_news_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
name: bert-base-german-cased
size_factor: 3
accuracy:
name: bert-base-german-cased
size_factor: 3
fr:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: fr_core_news_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
name: camembert-base
size_factor: 3
accuracy:
name: camembert-base
size_factor: 3
es:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: es_core_news_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
Update transformer recommendations [ci skip]
2020-10-13 13:41:17 +00:00
name: dccuchile/bert-base-spanish-wwm-cased
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
size_factor: 3
accuracy:
Update transformer recommendations [ci skip]
2020-10-13 13:41:17 +00:00
name: dccuchile/bert-base-spanish-wwm-cased
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
size_factor: 3
sv:
word_vectors: null
transformer:
efficiency:
name: KB/bert-base-swedish-cased
size_factor: 3
accuracy:
name: KB/bert-base-swedish-cased
size_factor: 3
fi:
word_vectors: null
transformer:
efficiency:
name: TurkuNLP/bert-base-finnish-cased-v1
size_factor: 3
accuracy:
name: TurkuNLP/bert-base-finnish-cased-v1
size_factor: 3
el:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: el_core_news_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
name: nlpaueb/bert-base-greek-uncased-v1
size_factor: 3
accuracy:
name: nlpaueb/bert-base-greek-uncased-v1
size_factor: 3
tr:
word_vectors: null
transformer:
efficiency:
name: dbmdz/bert-base-turkish-cased
size_factor: 3
accuracy:
name: dbmdz/bert-base-turkish-cased
size_factor: 3
zh:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: zh_core_web_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
name: bert-base-chinese
size_factor: 3
accuracy:
name: bert-base-chinese
size_factor: 3
has_letters: false
ar:
word_vectors: null
transformer:
efficiency:
name: asafaya/bert-base-arabic
size_factor: 3
accuracy:
name: asafaya/bert-base-arabic
size_factor: 3
pl:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: pl_core_news_lg
Update init config and recommendations - As much as I dislike YAML, it seemed like a better format here because it allows us to add comments if we want to explain the different recommendations - Don't include the generated JS in the repo by default and build it on the fly when running or deploying the site. This ensures it's always up to date. - Simplify jinja_to_js script and use fewer dependencies
2020-08-19 11:33:15 +00:00
transformer:
efficiency:
name: dkleczek/bert-base-polish-cased-v1
size_factor: 3
accuracy:
name: dkleczek/bert-base-polish-cased-v1
size_factor: 3
Update transformer recommendations [ci skip]
2020-10-13 13:41:17 +00:00
nl:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: nl_core_news_lg
Update transformer recommendations [ci skip]
2020-10-13 13:41:17 +00:00
transformer:
efficiency:
name: pdelobelle/robbert-v2-dutch-base
size_factor: 3
accuracy:
name: pdelobelle/robbert-v2-dutch-base
size_factor: 3
pt:
Use core lg models as vectors models in quickstart
2020-10-16 06:17:53 +00:00
word_vectors: pt_core_news_lg
Update transformer recommendations [ci skip]
2020-10-13 13:41:17 +00:00
transformer:
efficiency:
name: neuralmind/bert-base-portuguese-cased
size_factor: 3
accuracy:
name: neuralmind/bert-base-portuguese-cased
size_factor: 3