File size: 1,510 Bytes
5c93031 a2304a1 f32907e 458aedf 5c93031 f32907e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
import gradio as gr
import os
import csv
import time
token = os.getenv('HF_TOKEN')
def save_feedback(text, original_output, corrected_label):
with open('/data/feedback_data.csv', 'a', newline='') as f:
predicted_label = max(original_output, key=original_output.get)
writer = csv.writer(f)
writer.writerow([text, original_output, predicted_label, corrected_label])
return "Feedback saved successfully!"
def clear_message():
time.sleep(1) # Wait for 1 seconds
return gr.update(value="")
with gr.Blocks() as demo:
# Load the classification app
classification_app = gr.load("models/ressolve-ai/sentiment_cobranzas_BAL4", hf_token=token)
# Get the input and output components from the loaded app
text_input = classification_app.input_components[0]
output_label = classification_app.output_components[0]
# Add a subtitle before the feedback section
gr.Markdown("## Feedback the sentiment")
# Add feedback buttons below the classification app
with gr.Row():
feedback_label = gr.Radio(["Positivo", "Negativo", "Neutro"], label="Correct Label")
feedback_btn = gr.Button("If applicable, press this button to correct the label")
feedback_message = gr.Textbox(label="Feedback Status", interactive=False)
feedback_btn.click(save_feedback, inputs=[text_input, output_label, feedback_label], outputs=feedback_message)
feedback_btn.click(clear_message, inputs=None, outputs=feedback_message)
demo.launch() |