|
|
|
|
|
from transformers import pipeline |
|
|
import torch |
|
|
import time |
|
|
import gradio as gr |
|
|
|
|
|
|
|
|
text_model = pipeline( |
|
|
"text2text-generation", |
|
|
model="google/flan-t5-base", |
|
|
device=-1) |
|
|
emotion_model = pipeline( |
|
|
"text-classification", |
|
|
model="SamLowe/roberta-base-go_emotions", |
|
|
device=-1, |
|
|
return_all_scores=True |
|
|
) |
|
|
|
|
|
|
|
|
def summarize(text): |
|
|
summary = text_model( |
|
|
f"summarize: {text}", |
|
|
max_new_tokens=250, |
|
|
min_length=70, |
|
|
do_sample=True, |
|
|
top_k=50, |
|
|
top_p=0.95, |
|
|
temperature=0.9, |
|
|
num_beams=3 |
|
|
) |
|
|
return summary[0]['generated_text'] |
|
|
|
|
|
def answer_question(context, question): |
|
|
prompt = f"Answer the question based on the context:\nContext: {context}\nQuestion: {question}" |
|
|
answer = text_model(prompt, max_new_tokens=100) |
|
|
return answer[0]['generated_text'] |
|
|
|
|
|
def classify_emotion(sentence): |
|
|
emotions = emotion_model(sentence) |
|
|
top_emotion = max(emotions[0], key=lambda x: x['score']) |
|
|
return top_emotion['label'] |
|
|
|
|
|
|
|
|
def gr_summarize(text): |
|
|
return summarize(text) |
|
|
|
|
|
def gr_qa(context, question): |
|
|
return answer_question(context, question) |
|
|
|
|
|
def gr_emotion(sentence): |
|
|
return classify_emotion(sentence) |
|
|
|
|
|
|
|
|
with gr.Blocks(css=""" |
|
|
body {background-color: black; color: red;} |
|
|
.gr-button {background-color: red; color: black;} |
|
|
.gr-box {background-color: #111111; color: red;} |
|
|
.gr-textbox textarea {background-color: #111111; color: red;} |
|
|
.gr-tabs {background-color: #111111; color: red;} |
|
|
""") as chad_app: |
|
|
|
|
|
with gr.Tab("Homepage"): |
|
|
gr.Markdown(""" |
|
|
## Welcome to Chad-1: A Text Intelligence Assistant |
|
|
Developed by Olaleye Faithfulness Ibukun |
|
|
Use the tabs above to Summarize text, Ask a question, or Classify emotion. |
|
|
""") |
|
|
|
|
|
with gr.Tab("Summarize"): |
|
|
input_text = gr.Textbox(label="Enter text to summarize", placeholder="Enter the text you want summarized", lines=6) |
|
|
output_summary = gr.Textbox(label="Summary", lines=6) |
|
|
btn_summarize = gr.Button("Summarize") |
|
|
btn_summarize.click(fn=gr_summarize, inputs=input_text, outputs=output_summary) |
|
|
|
|
|
with gr.Tab("Ask a Question"): |
|
|
input_context = gr.Textbox(label="Context / Scenario", placeholder="Enter the context or scenario surrounding your question", lines=6) |
|
|
input_question = gr.Textbox(label="Question", placeholder="Now, ask a question based on the context above") |
|
|
output_answer = gr.Textbox(label="Answer", lines=6) |
|
|
btn_qa = gr.Button("Get Answer") |
|
|
btn_qa.click(fn=gr_qa, inputs=[input_context, input_question], outputs=output_answer) |
|
|
|
|
|
with gr.Tab("Classify Emotion"): |
|
|
input_sentence = gr.Textbox(label="Sentence", placeholder="Enter a sentence to get the underlying emotion", lines=2) |
|
|
output_emotion = gr.Textbox(label="Emotion") |
|
|
btn_emotion = gr.Button("Classify") |
|
|
btn_emotion.click(fn=gr_emotion, inputs=input_sentence, outputs=output_emotion) |
|
|
|
|
|
|
|
|
chad_app.launch(share=True) |
|
|
|
|
|
|
|
|
|
|
|
|
