Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
from sentence_transformers import SentenceTransformer, util
|
| 3 |
-
import
|
| 4 |
import os
|
| 5 |
|
| 6 |
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
|
@@ -9,17 +9,18 @@ os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
|
| 9 |
filename = "output_topic_details.txt" # Path to the file storing destress-specific details
|
| 10 |
retrieval_model_name = 'output/sentence-transformer-finetuned/'
|
| 11 |
|
| 12 |
-
#
|
|
|
|
|
|
|
| 13 |
|
| 14 |
-
system_message = "You are a comfort chatbot specialized in providing information on therapy, destressing
|
| 15 |
-
# Initial system message to set the behavior of the assistant
|
| 16 |
messages = [{"role": "system", "content": system_message}]
|
| 17 |
messages.append({
|
| 18 |
-
|
| 19 |
-
|
| 20 |
})
|
| 21 |
|
| 22 |
-
#
|
| 23 |
try:
|
| 24 |
retrieval_model = SentenceTransformer(retrieval_model_name)
|
| 25 |
print("Models loaded successfully.")
|
|
@@ -27,9 +28,6 @@ except Exception as e:
|
|
| 27 |
print(f"Failed to load models: {e}")
|
| 28 |
|
| 29 |
def load_and_preprocess_text(filename):
|
| 30 |
-
"""
|
| 31 |
-
Load and preprocess text from a file, removing empty lines and stripping whitespace.
|
| 32 |
-
"""
|
| 33 |
try:
|
| 34 |
with open(filename, 'r', encoding='utf-8') as file:
|
| 35 |
segments = [line.strip() for line in file if line.strip()]
|
|
@@ -42,56 +40,29 @@ def load_and_preprocess_text(filename):
|
|
| 42 |
segments = load_and_preprocess_text(filename)
|
| 43 |
|
| 44 |
def find_relevant_segment(user_query, segments):
|
| 45 |
-
"""
|
| 46 |
-
Find the most relevant text segment for a user's query using cosine similarity among sentence embeddings.
|
| 47 |
-
This version finds the best match based on the content of the query.
|
| 48 |
-
"""
|
| 49 |
try:
|
| 50 |
-
# Lowercase the query for better matching
|
| 51 |
lower_query = user_query.lower()
|
| 52 |
-
|
| 53 |
-
# Encode the query and the segments
|
| 54 |
query_embedding = retrieval_model.encode(lower_query)
|
| 55 |
segment_embeddings = retrieval_model.encode(segments)
|
| 56 |
-
|
| 57 |
-
# Compute cosine similarities between the query and the segments
|
| 58 |
similarities = util.pytorch_cos_sim(query_embedding, segment_embeddings)[0]
|
| 59 |
-
|
| 60 |
-
# Find the index of the most similar segment
|
| 61 |
best_idx = similarities.argmax()
|
| 62 |
-
|
| 63 |
-
# Return the most relevant segment
|
| 64 |
return segments[best_idx]
|
| 65 |
except Exception as e:
|
| 66 |
print(f"Error in finding relevant segment: {e}")
|
| 67 |
return ""
|
| 68 |
|
| 69 |
def generate_response(user_query, relevant_segment):
|
| 70 |
-
"""
|
| 71 |
-
Generate a response emphasizing the bot's capability in providing therapy, destressing activites, and student opportunities information.
|
| 72 |
-
"""
|
| 73 |
try:
|
| 74 |
user_message = f"Here's the information on your request: {relevant_segment}"
|
| 75 |
-
|
| 76 |
-
# Append user's message to messages list
|
| 77 |
messages.append({"role": "user", "content": user_message})
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
top_p=1,
|
| 85 |
-
frequency_penalty=0.5,
|
| 86 |
-
presence_penalty=0.5,
|
| 87 |
-
)
|
| 88 |
-
|
| 89 |
-
# Extract the response text
|
| 90 |
-
output_text = response['choices'][0]['message']['content'].strip()
|
| 91 |
-
|
| 92 |
# Append assistant's message to messages list for context
|
| 93 |
messages.append({"role": "assistant", "content": output_text})
|
| 94 |
-
|
| 95 |
return output_text
|
| 96 |
|
| 97 |
except Exception as e:
|
|
@@ -99,9 +70,6 @@ def generate_response(user_query, relevant_segment):
|
|
| 99 |
return f"Error in generating response: {e}"
|
| 100 |
|
| 101 |
def query_model(question):
|
| 102 |
-
"""
|
| 103 |
-
Process a question, find relevant information, and generate a response.
|
| 104 |
-
"""
|
| 105 |
if question == "":
|
| 106 |
return "Welcome to CalmConnect! Ask me anything about destressing strategies or student opportunities. Feel free to talk to our online therapist!"
|
| 107 |
relevant_segment = find_relevant_segment(question, segments)
|
|
@@ -111,112 +79,33 @@ def query_model(question):
|
|
| 111 |
return response
|
| 112 |
|
| 113 |
# Define the HTML iframe content
|
| 114 |
-
iframe
|
| 115 |
-
<iframe style="border-radius:12px" src="https://docs.google.com/spreadsheets/d/e/2PACX-1vRroWVBXq1Fa0x7SvRTzSBMHFIp59VtVEWCxeg8kWJU4ll1_o4yzBnt4ArT88s7g4TQrMKEXZUQAeHF/pubhtml?widget=true&headers=false" width="100%" height="352" frameBorder="0" allowfullscreen="true" allow="autoplay; clipboard-write; encrypted-media; fullscreen; picture-in-picture" loading="lazy"></iframe>
|
| 116 |
-
'''
|
| 117 |
-
|
| 118 |
-
iframe2 = '''
|
| 119 |
-
<iframe style="border-radius:12px" src="https://open.spotify.com/embed/playlist/6wwxTePuIKYMqt6RCytB7X?utm_source=generator" width="100%" height="300" frameBorder="0" allowfullscreen="" allow="autoplay; clipboard-write; encrypted-media; fullscreen; picture-in-picture" loading="lazy"></iframe>
|
| 120 |
-
'''
|
| 121 |
-
|
| 122 |
|
| 123 |
# Define the welcome message and specific topics the chatbot can provide information about
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
"""
|
| 127 |
-
|
| 128 |
-
"""
|
| 129 |
-
## Your AI-driven assistant for destressing and extracurricular opportunity queries. Created by Olivia W, Alice T, and Cindy W of the 2024 Kode With Klossy CITY Camp.
|
| 130 |
-
"""
|
| 131 |
-
|
| 132 |
-
topics = """
|
| 133 |
-
### If you are interested in the following below, click on our Student Opportunities Database!
|
| 134 |
-
- Engineering
|
| 135 |
-
- Technology / Computer Science
|
| 136 |
-
- Research : STEM
|
| 137 |
-
- Finance
|
| 138 |
-
- Law / Political Science / Debate
|
| 139 |
-
- The Arts
|
| 140 |
-
- Business / Leadership
|
| 141 |
-
- Pyschology
|
| 142 |
-
- Medicine / Biology
|
| 143 |
-
- Literature / Writing
|
| 144 |
-
- College Prep
|
| 145 |
-
- Advocacy: Non-Profit, Environment or Identity
|
| 146 |
-
- Volunteering
|
| 147 |
-
- Study Abroad
|
| 148 |
-
"""
|
| 149 |
-
|
| 150 |
-
topics2= """
|
| 151 |
-
### Feel Free to ask CalmBot (Our Therapist Bot) anything from the topics below!
|
| 152 |
-
- Arts and Crafts (When asking for arts and crafts ideas, state whether you have 15 min, 30 min, 45 min, 1 hour, 1 hour and a half, 2 hours, 2 hours and a half, 3 hours or greater)
|
| 153 |
-
- Destressing strategies (Breathing Exercises, stretches, etc.)
|
| 154 |
-
- Mental Health
|
| 155 |
-
- Identity (Sexual, Gender, etc.)
|
| 156 |
-
- Bullying
|
| 157 |
-
- Racism
|
| 158 |
-
- Relationships (Family, Friends, etc.)
|
| 159 |
-
- Abuse (Emotional, Physical, Sexual, Mental, etc.)
|
| 160 |
-
- Support Resources
|
| 161 |
-
"""
|
| 162 |
-
|
| 163 |
-
# Create a Gradio HTML component
|
| 164 |
-
def display_iframe():
|
| 165 |
-
return iframe
|
| 166 |
-
|
| 167 |
-
def display_iframe2():
|
| 168 |
-
return iframe2
|
| 169 |
-
|
| 170 |
-
|
| 171 |
-
theme = gr.themes.Default(
|
| 172 |
-
primary_hue="neutral",
|
| 173 |
-
secondary_hue="neutral",
|
| 174 |
-
).set(
|
| 175 |
-
background_fill_primary='#e3e9da',
|
| 176 |
-
background_fill_primary_dark='#e3e9da',
|
| 177 |
-
background_fill_secondary="#f8f1ea",
|
| 178 |
-
background_fill_secondary_dark="#f8f1ea",
|
| 179 |
-
border_color_accent="#f8f1ea",
|
| 180 |
-
border_color_accent_dark="#e3e9da",
|
| 181 |
-
border_color_accent_subdued="#f8f1ea",
|
| 182 |
-
border_color_primary="#f8f1ea",
|
| 183 |
-
block_border_color="#f8f1ea",
|
| 184 |
-
button_primary_background_fill="#f8f1ea",
|
| 185 |
-
button_primary_background_fill_dark="#f8f1ea"
|
| 186 |
-
)
|
| 187 |
-
|
| 188 |
# Setup the Gradio Blocks interface with custom layout components
|
| 189 |
-
with gr.Blocks(
|
| 190 |
gr.Image("CalmConnect.jpg", show_label=False, show_share_button=False, show_download_button=False)
|
| 191 |
-
gr.Markdown(welcome_message)
|
| 192 |
-
|
| 193 |
with gr.Row():
|
| 194 |
with gr.Column():
|
| 195 |
-
gr.Markdown(topics)
|
| 196 |
-
gr.HTML(iframe)
|
| 197 |
-
gr.HTML(iframe2)
|
| 198 |
|
| 199 |
with gr.Column():
|
| 200 |
-
gr.Markdown(topics2)
|
| 201 |
with gr.Row():
|
| 202 |
with gr.Column():
|
| 203 |
question = gr.Textbox(label="You", placeholder="What do you want to talk to CalmBot about?")
|
| 204 |
answer = gr.Textbox(label="CalmBot's Response :D", placeholder="CalmBot will respond here..", interactive=False, lines=20)
|
| 205 |
submit_button = gr.Button("Submit")
|
| 206 |
submit_button.click(fn=query_model, inputs=question, outputs=answer)
|
| 207 |
-
|
| 208 |
-
with gr.Row():
|
| 209 |
-
big_block = gr.HTML("### <button><a href='https://www.headspace.com/teens'>FREE: HEADSPACE FOR TEENS </a></button>")
|
| 210 |
-
big_block2 = gr.HTML("<button><a href='https://calmconnect-flower.replit.app/'>PLAY FLOWER GAME</a></button>")
|
| 211 |
-
big_block3 = gr.HTML("<button><a href='https://www.nyc.gov/site/doh/health/health-topics/teenspace.page'>NYC: TEENSPACE (free services)</a></button>")
|
| 212 |
-
big_block4 =gr.HTML("<button><a href='https://www.teenlife.com/blog/mental-health-resources-for-teens/'>TEEN MENTAL HEALTH RESOURCES (free services)</a></button>")
|
| 213 |
-
|
| 214 |
-
|
| 215 |
-
demo.launch()
|
| 216 |
-
|
| 217 |
-
|
| 218 |
-
|
| 219 |
|
|
|
|
|
|
|
| 220 |
|
| 221 |
# Launch the Gradio app to allow user interaction
|
| 222 |
-
demo.launch(share=True)
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
from sentence_transformers import SentenceTransformer, util
|
| 3 |
+
from transformers import GPT2LMHeadModel, GPT2Tokenizer
|
| 4 |
import os
|
| 5 |
|
| 6 |
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
|
|
|
| 9 |
filename = "output_topic_details.txt" # Path to the file storing destress-specific details
|
| 10 |
retrieval_model_name = 'output/sentence-transformer-finetuned/'
|
| 11 |
|
| 12 |
+
# Load GPT-2 model and tokenizer
|
| 13 |
+
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
|
| 14 |
+
model = GPT2LMHeadModel.from_pretrained("gpt2")
|
| 15 |
|
| 16 |
+
system_message = "You are a comfort chatbot specialized in providing information on therapy, destressing activities, and student opportunities."
|
|
|
|
| 17 |
messages = [{"role": "system", "content": system_message}]
|
| 18 |
messages.append({
|
| 19 |
+
"role": "system",
|
| 20 |
+
"content": "Do not use Markdown Format. Do not include hashtags or asterisks"
|
| 21 |
})
|
| 22 |
|
| 23 |
+
# Load the retrieval model
|
| 24 |
try:
|
| 25 |
retrieval_model = SentenceTransformer(retrieval_model_name)
|
| 26 |
print("Models loaded successfully.")
|
|
|
|
| 28 |
print(f"Failed to load models: {e}")
|
| 29 |
|
| 30 |
def load_and_preprocess_text(filename):
|
|
|
|
|
|
|
|
|
|
| 31 |
try:
|
| 32 |
with open(filename, 'r', encoding='utf-8') as file:
|
| 33 |
segments = [line.strip() for line in file if line.strip()]
|
|
|
|
| 40 |
segments = load_and_preprocess_text(filename)
|
| 41 |
|
| 42 |
def find_relevant_segment(user_query, segments):
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
try:
|
|
|
|
| 44 |
lower_query = user_query.lower()
|
|
|
|
|
|
|
| 45 |
query_embedding = retrieval_model.encode(lower_query)
|
| 46 |
segment_embeddings = retrieval_model.encode(segments)
|
|
|
|
|
|
|
| 47 |
similarities = util.pytorch_cos_sim(query_embedding, segment_embeddings)[0]
|
|
|
|
|
|
|
| 48 |
best_idx = similarities.argmax()
|
|
|
|
|
|
|
| 49 |
return segments[best_idx]
|
| 50 |
except Exception as e:
|
| 51 |
print(f"Error in finding relevant segment: {e}")
|
| 52 |
return ""
|
| 53 |
|
| 54 |
def generate_response(user_query, relevant_segment):
|
|
|
|
|
|
|
|
|
|
| 55 |
try:
|
| 56 |
user_message = f"Here's the information on your request: {relevant_segment}"
|
|
|
|
|
|
|
| 57 |
messages.append({"role": "user", "content": user_message})
|
| 58 |
+
|
| 59 |
+
# Encode the input and generate a response
|
| 60 |
+
input_ids = tokenizer.encode(user_message, return_tensors='pt')
|
| 61 |
+
output = model.generate(input_ids, max_length=150, num_return_sequences=1)
|
| 62 |
+
output_text = tokenizer.decode(output[0], skip_special_tokens=True)
|
| 63 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
# Append assistant's message to messages list for context
|
| 65 |
messages.append({"role": "assistant", "content": output_text})
|
|
|
|
| 66 |
return output_text
|
| 67 |
|
| 68 |
except Exception as e:
|
|
|
|
| 70 |
return f"Error in generating response: {e}"
|
| 71 |
|
| 72 |
def query_model(question):
|
|
|
|
|
|
|
|
|
|
| 73 |
if question == "":
|
| 74 |
return "Welcome to CalmConnect! Ask me anything about destressing strategies or student opportunities. Feel free to talk to our online therapist!"
|
| 75 |
relevant_segment = find_relevant_segment(question, segments)
|
|
|
|
| 79 |
return response
|
| 80 |
|
| 81 |
# Define the HTML iframe content
|
| 82 |
+
# (Your iframe content goes here)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
|
| 84 |
# Define the welcome message and specific topics the chatbot can provide information about
|
| 85 |
+
# (Your welcome message and topics go here)
|
| 86 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
# Setup the Gradio Blocks interface with custom layout components
|
| 88 |
+
with gr.Blocks() as demo:
|
| 89 |
gr.Image("CalmConnect.jpg", show_label=False, show_share_button=False, show_download_button=False)
|
| 90 |
+
gr.Markdown(welcome_message)
|
| 91 |
+
|
| 92 |
with gr.Row():
|
| 93 |
with gr.Column():
|
| 94 |
+
gr.Markdown(topics)
|
| 95 |
+
gr.HTML(iframe)
|
| 96 |
+
gr.HTML(iframe2)
|
| 97 |
|
| 98 |
with gr.Column():
|
| 99 |
+
gr.Markdown(topics2)
|
| 100 |
with gr.Row():
|
| 101 |
with gr.Column():
|
| 102 |
question = gr.Textbox(label="You", placeholder="What do you want to talk to CalmBot about?")
|
| 103 |
answer = gr.Textbox(label="CalmBot's Response :D", placeholder="CalmBot will respond here..", interactive=False, lines=20)
|
| 104 |
submit_button = gr.Button("Submit")
|
| 105 |
submit_button.click(fn=query_model, inputs=question, outputs=answer)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 106 |
|
| 107 |
+
with gr.Row():
|
| 108 |
+
# (Your buttons go here)
|
| 109 |
|
| 110 |
# Launch the Gradio app to allow user interaction
|
| 111 |
+
demo.launch(share=True)
|