nllb

Softcatala · Aug 15, 2024 · f414a4b · f414a4b
1 parent 81ea7fe
commit f414a4b
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 19 deletions.
diff --git a/languages/eng-cat/corpus.yml b/languages/eng-cat/corpus.yml
@@ -1,20 +1,16 @@
 source_files:
-    - corpus-raw/aina-eng-cat-1.en
-    - corpus-raw/aina-eng-cat-2.en
-    - corpus-raw/aina-eng-cat-3.en
-    - corpus-raw/aina-eng-cat-4.en
-    - corpus-raw/aina-eng-cat-5.en
-    - corpus-raw/aina-eng-cat-6.en
-    - corpus-raw/aina-eng-cat-7.en
-    - corpus-raw/aina-eng-cat-8.en
+    - corpus-raw/nllb-eng-cat-1.en
+    - corpus-raw/nllb-eng-cat-2.en
+    - corpus-raw/nllb-eng-cat-3.en
+    - corpus-raw/nllb-eng-cat-4.en
+    - corpus-raw/nllb-eng-cat-5.en
+    - corpus-raw/nllb-eng-cat-6.en
 
 target_files:
-    - corpus-raw/aina-eng-cat-1.ca
-    - corpus-raw/aina-eng-cat-2.ca
-    - corpus-raw/aina-eng-cat-3.ca
-    - corpus-raw/aina-eng-cat-4.ca
-    - corpus-raw/aina-eng-cat-5.ca
-    - corpus-raw/aina-eng-cat-6.ca
-    - corpus-raw/aina-eng-cat-7.ca
-    - corpus-raw/aina-eng-cat-8.ca
+    - corpus-raw/nllb-eng-cat-1.ca
+    - corpus-raw/nllb-eng-cat-2.ca
+    - corpus-raw/nllb-eng-cat-3.ca
+    - corpus-raw/nllb-eng-cat-4.ca
+    - corpus-raw/nllb-eng-cat-5.ca
+    - corpus-raw/nllb-eng-cat-6.ca
 
diff --git a/languages/get-corpuses.sh b/languages/get-corpuses.sh
@@ -1,7 +1,7 @@
 #!/bin/bash
 
 #git clone https://github.com/Softcatala/parallel-catalan-corpus corpus-raw
-git clone --single-branch --branch eng-cat-aina --depth 1 https://github.com/Softcatala/parallel-catalan-corpus corpus-raw
+git clone --single-branch --branch eng-cat-nllb --depth 1 https://github.com/Softcatala/parallel-catalan-corpus corpus-raw
 cd corpus-raw
 
 declare -a arr=("eng-cat" "deu-cat" "ita-cat" "fra-cat" "spa-cat" "nld-cat" "por-cat" "jpn-cat" "glg-cat" "oci-cat" "eus-cat")

diff --git a/languages/train-all-to-cat.sh b/languages/train-all-to-cat.sh
@@ -1,13 +1,13 @@
 #!/bin/bash
 
 declare -a arr=("eng-cat" "deu-cat" "fra-cat" "ita-cat" "spa-cat" "por-cat" "nld-cat" "jpn-cat" "glg-cat" "oci-cat" "eus-cat")
-#declare -a arr=("eng-cat")
+declare -a arr=("eng-cat")
 
 for dirname in "${arr[@]}"; do
     echo Processing $dirname
     pushd $dirname
     # Done at pre-process-all.sh
-    #./preprocess.sh
+    ./preprocess.sh
     rm -r -f run/
     ./voc.sh
     ./train.sh