update files to get annotated data

Author: thjbdvlt

README.md

@@ -7,11 +7,8 @@ solipCysme
 | --- | --- |
 | __Language__ | french |
 | __Name__ | `fr_solipcysme` |
-| __Version__ | `0.2.5` |
-| __spaCy__ | `==3.8.4` |
 | __Default Pipeline__ | `jusqucy_tokenizer`,`commecy_normalizer`, `jusqucy_normalizer`, `pretagger_hunspell`,`morphologizer`, `viceverser_lemmatizer`, `parser` |
 | __Components__ | [jusqucy_tokenizer](https://github.com/thjbdvlt/jusquci), [jusqucy_normalizer](https://github.com/thjbdvlt/jusquci), [commecy_normalizer](https://github.com/thjbdvlt/commecy), `morphologizer`, [viceverser_lemmatizer](https://github.com/thjbdvlt/spacy-viceverser), `parser` |
-| __Vectors__ | 669785 keys, 6697856 unique vectors (100 dimensions) |
 | __Sources__ | Corpus [narraFEATS](https://github.com/thjbdvlt/corpus-narraFEATS) (morphologizer), [Universal Dependencies](https://universaldependencies.org/fr/) (parser), [french-word-vectors](https://github.com/thjbdvlt/french-word-vectors) (vectors)|
 | __License__ | [GPL](https://www.gnu.org/licenses/gpl-3.0.html) |
 | __Author__ | [thjbdvlt](https://github.com/thjbdvlt) |

make_pipeline/morphologizer/.gitignore

@@ -1 +1 @@
-narrafeats
+narrafeats*

make_pipeline/morphologizer/get_data.sh

@@ -28,8 +28,8 @@ unzip() {
     untar
 }
 download() {
-    local version=v0.1.1
-    wget https://github.com/thjbdvlt/corpus-narraFEATS/releases/download/$version/narrafeats.tar.gz
+    # wget https://github.com/thjbdvlt/corpus-narraFEATS/releases/download/$version/narrafeats.tar.gz
+    wget https://github.com/thjbdvlt/corpus-narraFEATS/releases/latest/download/narrafeats.tar.gz
     unzip
 }
 

make_pipeline/morphologizer/train.sh

@@ -38,12 +38,14 @@ case "$size" in
     sm | md)
         width=128
         depth=3
-        rows=[2000,500,1000,2000]
+        features='["NORM","SENT_START"]'
+        rows=[2000,500,500,1000,2000]
         static=false;;
     lg)
         width=256
         depth=4
-        rows=[4000,1000,2000,4000]
+        features='["NORM","SENT_START"]'
+        rows=[2000,1000,1000,2000,4000]
         static=true;;
     *)
         echo "Unknown value for -s: $size" >&2
@@ -89,6 +91,7 @@ opts+=(
     --paths.init_tok2vec=${pretrain_model}
     --${encode}.width=${width}
     --${encode}.depth=${depth}
+    --${embed}.features=${features}
     --${embed}.rows=${rows}
     --${embed}.include_static_vectors=${static}
     --paths.dev=${dev}

make_pipeline/package.sh

@@ -26,7 +26,7 @@ optional:
 size=
 
 # Default values
-name=solipCysme
+name=solipcysme
 output=pipeline
 meta=meta.json
 raw=data/raw.txt