Spaces:
Sleeping
Sleeping
Remove low quality languages from target languages (#5)
Browse files- Remove low quality languages from target languages (49900d0cf969370e83745811c86ce1d8ed17b1db)
app.py
CHANGED
|
@@ -8,12 +8,15 @@ import nltk
|
|
| 8 |
|
| 9 |
nltk.download("punkt")
|
| 10 |
|
|
|
|
|
|
|
|
|
|
| 11 |
device = "cpu" if platform.system() == "Darwin" else "cuda"
|
| 12 |
MODEL_NAME = "facebook/nllb-200-3.3B"
|
| 13 |
|
| 14 |
code_mapping = dict(sorted(code_mapping.items(), key=lambda item: item[1]))
|
| 15 |
flores_codes = list(code_mapping.keys())
|
| 16 |
-
|
| 17 |
|
| 18 |
def load_model():
|
| 19 |
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME).to(device)
|
|
@@ -77,7 +80,7 @@ with gr.Blocks() as demo:
|
|
| 77 |
gr.Markdown(description)
|
| 78 |
with gr.Row():
|
| 79 |
src_lang = gr.Dropdown(label="Source Language", choices=flores_codes)
|
| 80 |
-
target_lang = gr.Dropdown(label="Target Language", choices=
|
| 81 |
with gr.Row():
|
| 82 |
input_text = gr.Textbox(label="Input Text", lines=6)
|
| 83 |
with gr.Row():
|
|
|
|
| 8 |
|
| 9 |
nltk.download("punkt")
|
| 10 |
|
| 11 |
+
REMOVED_TARGET_LANGUAGES = {"Ligurian", "Lombard", "Sicilian"}
|
| 12 |
+
|
| 13 |
+
|
| 14 |
device = "cpu" if platform.system() == "Darwin" else "cuda"
|
| 15 |
MODEL_NAME = "facebook/nllb-200-3.3B"
|
| 16 |
|
| 17 |
code_mapping = dict(sorted(code_mapping.items(), key=lambda item: item[1]))
|
| 18 |
flores_codes = list(code_mapping.keys())
|
| 19 |
+
target_languages = [language for language in flores_codes if not language in REMOVED_TARGET_LANGUAGES]
|
| 20 |
|
| 21 |
def load_model():
|
| 22 |
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME).to(device)
|
|
|
|
| 80 |
gr.Markdown(description)
|
| 81 |
with gr.Row():
|
| 82 |
src_lang = gr.Dropdown(label="Source Language", choices=flores_codes)
|
| 83 |
+
target_lang = gr.Dropdown(label="Target Language", choices=target_languages)
|
| 84 |
with gr.Row():
|
| 85 |
input_text = gr.Textbox(label="Input Text", lines=6)
|
| 86 |
with gr.Row():
|