Spaces:
Running
Running
sabarinathan
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,3 @@
|
|
1 |
-
|
2 |
# -*- coding: utf-8 -*-
|
3 |
"""
|
4 |
Created on Tue Sep 17 19:03:17 2024
|
@@ -53,7 +52,7 @@ class JapaneseNER():
|
|
53 |
sample_encoding = self.tokenizer([
|
54 |
"鈴木は4月の陽気の良い日に、鈴をつけて熊本県の阿蘇山に登った",
|
55 |
"中国では、中国共産党による一党統治が続く",
|
56 |
-
], truncation=True,padding=True, # Ensure all sequences are of the same length
|
57 |
max_length=512, return_tensors="pt")
|
58 |
|
59 |
sample_encoding = {k: v.to(device) for k, v in sample_encoding.items()}
|
@@ -66,7 +65,7 @@ class JapaneseNER():
|
|
66 |
print("Predicted labels:", predicted_label_id)
|
67 |
|
68 |
def predict(self, text):
|
69 |
-
encoding = self.tokenizer([text], truncation=True,padding=True, max_length=512, return_tensors="pt")
|
70 |
encoding = {k: v.to(device) for k, v in encoding.items()}
|
71 |
|
72 |
# Perform prediction
|
@@ -106,16 +105,19 @@ def ner_inference(text):
|
|
106 |
start_idx += 1
|
107 |
doc.ents = ents # Set the entities in the Doc
|
108 |
|
109 |
-
# Render using
|
110 |
html = displacy.render(doc, style="ent", jupyter=False) # Generate HTML for entities
|
111 |
return html
|
112 |
|
|
|
|
|
|
|
113 |
# Create Gradio interface
|
114 |
import gradio as gr
|
115 |
|
116 |
iface = gr.Interface(
|
117 |
fn=ner_inference, # The function to call for prediction
|
118 |
-
inputs=gr.Textbox(lines=5, placeholder="Enter Japanese text for NER..."), # Input widget
|
119 |
outputs="html", # Output will be in HTML format using displacy
|
120 |
title="Japanese Named Entity Recognition (NER)",
|
121 |
description="Enter Japanese text and see the named entities highlighted in the output."
|
|
|
|
|
1 |
# -*- coding: utf-8 -*-
|
2 |
"""
|
3 |
Created on Tue Sep 17 19:03:17 2024
|
|
|
52 |
sample_encoding = self.tokenizer([
|
53 |
"鈴木は4月の陽気の良い日に、鈴をつけて熊本県の阿蘇山に登った",
|
54 |
"中国では、中国共産党による一党統治が続く",
|
55 |
+
], truncation=True, padding=True, # Ensure all sequences are of the same length
|
56 |
max_length=512, return_tensors="pt")
|
57 |
|
58 |
sample_encoding = {k: v.to(device) for k, v in sample_encoding.items()}
|
|
|
65 |
print("Predicted labels:", predicted_label_id)
|
66 |
|
67 |
def predict(self, text):
|
68 |
+
encoding = self.tokenizer([text], truncation=True, padding=True, max_length=512, return_tensors="pt")
|
69 |
encoding = {k: v.to(device) for k, v in encoding.items()}
|
70 |
|
71 |
# Perform prediction
|
|
|
105 |
start_idx += 1
|
106 |
doc.ents = ents # Set the entities in the Doc
|
107 |
|
108 |
+
# Render using spaCy displacy
|
109 |
html = displacy.render(doc, style="ent", jupyter=False) # Generate HTML for entities
|
110 |
return html
|
111 |
|
112 |
+
# Sample text for demonstration
|
113 |
+
sample_text = "鈴木一朗は2020年に引退した。女優の石原さとみは多くの映画で主演している。"
|
114 |
+
|
115 |
# Create Gradio interface
|
116 |
import gradio as gr
|
117 |
|
118 |
iface = gr.Interface(
|
119 |
fn=ner_inference, # The function to call for prediction
|
120 |
+
inputs=gr.Textbox(lines=5, placeholder="Enter Japanese text for NER...", default=sample_text), # Input widget with sample text
|
121 |
outputs="html", # Output will be in HTML format using displacy
|
122 |
title="Japanese Named Entity Recognition (NER)",
|
123 |
description="Enter Japanese text and see the named entities highlighted in the output."
|