Spaces:
Runtime error
Runtime error
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
import transformers as trf
|
4 |
+
|
5 |
+
# Load the summarization model
|
6 |
+
summarization_model_path = 'ieuniversity/News-Translator'
|
7 |
+
summarization_tokenizer = trf.AutoTokenizer.from_pretrained(summarization_model_path)
|
8 |
+
summarization_model = trf.AutoModelForSeq2SeqLM.from_pretrained(summarization_model_path)
|
9 |
+
|
10 |
+
# Load the translation model
|
11 |
+
translation_model_path = 'hackathon-pln-es/t5-small-finetuned-spanish-to-quechua'
|
12 |
+
translation_tokenizer = trf.AutoTokenizer.from_pretrained(translation_model_path)
|
13 |
+
translation_model = trf.AutoModelForSeq2SeqLM.from_pretrained(translation_model_path)
|
14 |
+
|
15 |
+
def summarize_and_translate(news_text):
|
16 |
+
# Summarize the news article
|
17 |
+
max_input_length = 512
|
18 |
+
max_output_length = 128
|
19 |
+
input_encoded = summarization_tokenizer.encode_plus(news_text, add_special_tokens=True,
|
20 |
+
max_length=max_input_length, pad_to_max_length=True,
|
21 |
+
return_attention_mask=True, return_tensors='pt')
|
22 |
+
input_ids = input_encoded['input_ids']
|
23 |
+
attention_mask = input_encoded['attention_mask']
|
24 |
+
output_ids = summarization_model.generate(input_ids=input_ids, attention_mask=attention_mask,
|
25 |
+
max_length=max_output_length)
|
26 |
+
summary_text = summarization_tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
27 |
+
|
28 |
+
# Translate the summary to Quechua
|
29 |
+
input_encoded = translation_tokenizer(summary_text, padding=True, truncation=True, max_length=512, return_tensors='pt')
|
30 |
+
input_ids = input_encoded['input_ids']
|
31 |
+
attention_mask = input_encoded['attention_mask']
|
32 |
+
output_ids = translation_model.generate(input_ids=input_ids, attention_mask=attention_mask,
|
33 |
+
max_length=512)
|
34 |
+
output_text = translation_tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
35 |
+
|
36 |
+
return output_text
|
37 |
+
|
38 |
+
# Define the input and output interfaces for Gradio
|
39 |
+
input_interface = gr.inputs.Textbox(label="Input News Text (Spanish)")
|
40 |
+
output_interface = gr.outputs.Textbox(label="Output News Text (Quechua)")
|
41 |
+
|
42 |
+
# Create and launch the Gradio app
|
43 |
+
|
44 |
+
iface = gr.Interface(fn=summarize_and_translate, inputs=input_interface, outputs=output_interface, title="Spanish-to-Quechua News Summarization and Translation")
|
45 |
+
iface.launch()
|