scholar-2001
commited on
Commit
β’
4cfc34d
1
Parent(s):
9db63ff
Final
Browse files- .gitattributes +3 -0
- app.py +239 -0
- chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/data_level0.bin +3 -0
- chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/header.bin +3 -0
- chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/length.bin +3 -0
- chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/link_lists.bin +0 -0
- chroma_db/chroma.sqlite3 +3 -0
- requiremnts.txt +3 -0
.gitattributes
CHANGED
@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
chromadb/*.bin filter=lfs diff=lfs merge=lfs -text
|
37 |
+
chromadb/*.sqlite3 filter=lfs diff=lfs merge=lfs -text
|
38 |
+
chroma_db/**/* filter=lfs diff=lfs merge=lfs -text
|
app.py
ADDED
@@ -0,0 +1,239 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import chromadb
|
3 |
+
from chromadb.utils import embedding_functions
|
4 |
+
import groq
|
5 |
+
from typing import Dict
|
6 |
+
import os
|
7 |
+
|
8 |
+
class CourseAdvisor:
|
9 |
+
def __init__(self, db_path: str = "./chroma_db"):
|
10 |
+
"""Initialize the course advisor with existing ChromaDB database."""
|
11 |
+
# Initialize persistent client with path
|
12 |
+
self.chroma_client = chromadb.PersistentClient(path=db_path)
|
13 |
+
|
14 |
+
# Initialize embedding function
|
15 |
+
self.embedding_function = embedding_functions.SentenceTransformerEmbeddingFunction(
|
16 |
+
model_name="jinaai/jina-embeddings-v2-base-en"
|
17 |
+
)
|
18 |
+
|
19 |
+
# Get existing collection
|
20 |
+
self.collection = self.chroma_client.get_collection(
|
21 |
+
name="courses",
|
22 |
+
embedding_function=self.embedding_function
|
23 |
+
)
|
24 |
+
|
25 |
+
def query_courses(self, query_text: str, chat_history: str, api_key: str, n_results: int = 3) -> Dict:
|
26 |
+
"""Query the vector database and get course recommendations."""
|
27 |
+
# Initialize Groq client with provided API key
|
28 |
+
groq_client = groq.Groq(api_key=api_key)
|
29 |
+
|
30 |
+
try:
|
31 |
+
# Get relevant documents from vector DB
|
32 |
+
results = self.collection.query(
|
33 |
+
query_texts=[query_text],
|
34 |
+
n_results=min(n_results, self.collection.count()),
|
35 |
+
include=['documents', 'metadatas']
|
36 |
+
)
|
37 |
+
|
38 |
+
# Prepare context from retrieved documents
|
39 |
+
docs_context = "\n\n".join(results['documents'][0])
|
40 |
+
|
41 |
+
except Exception as e:
|
42 |
+
st.error(f"Error querying database: {str(e)}")
|
43 |
+
return {
|
44 |
+
'llm_response': "I encountered an error while searching the course database. Please try again.",
|
45 |
+
'retrieved_courses': []
|
46 |
+
}
|
47 |
+
|
48 |
+
# Create prompt with chat history
|
49 |
+
prompt = f"""Previous conversation:
|
50 |
+
{chat_history}
|
51 |
+
|
52 |
+
Current user query: {query_text}
|
53 |
+
|
54 |
+
Relevant course information:
|
55 |
+
{docs_context}
|
56 |
+
|
57 |
+
Please provide course recommendations based on the entire conversation context. Format your response as:
|
58 |
+
1. Understanding of the user's needs (based on conversation history)
|
59 |
+
2. Overall recommendation with reasoning
|
60 |
+
3. Specific benefits of each recommended course
|
61 |
+
4. Learning path suggestion (if applicable)
|
62 |
+
5. Any prerequisites or important notes"""
|
63 |
+
|
64 |
+
try:
|
65 |
+
# Get response from Groq
|
66 |
+
completion = groq_client.chat.completions.create(
|
67 |
+
messages=[
|
68 |
+
{"role": "system", "content": "You are a helpful course advisor who provides detailed, relevant course recommendations based on the user's needs and conversation history. Keep responses clear and well-structured."},
|
69 |
+
{"role": "user", "content": prompt}
|
70 |
+
],
|
71 |
+
model="mixtral-8x7b-32768",
|
72 |
+
temperature=0.7,
|
73 |
+
)
|
74 |
+
|
75 |
+
return {
|
76 |
+
'llm_response': completion.choices[0].message.content,
|
77 |
+
'retrieved_courses': results['metadatas'][0]
|
78 |
+
}
|
79 |
+
|
80 |
+
except Exception as e:
|
81 |
+
st.error(f"Error with Groq API: {str(e)}")
|
82 |
+
return {
|
83 |
+
'llm_response': "I encountered an error while generating recommendations. Please check your API key and try again.",
|
84 |
+
'retrieved_courses': []
|
85 |
+
}
|
86 |
+
|
87 |
+
def initialize_session_state():
|
88 |
+
"""Initialize session state variables."""
|
89 |
+
if 'messages' not in st.session_state:
|
90 |
+
st.session_state.messages = []
|
91 |
+
if 'course_advisor' not in st.session_state:
|
92 |
+
st.session_state.course_advisor = CourseAdvisor()
|
93 |
+
if 'api_key' not in st.session_state:
|
94 |
+
st.session_state.api_key = ""
|
95 |
+
|
96 |
+
def get_chat_history() -> str:
|
97 |
+
"""Format chat history for LLM context."""
|
98 |
+
history = []
|
99 |
+
for message in st.session_state.messages[-5:]: # Only use last 5 messages for context
|
100 |
+
role = message["role"]
|
101 |
+
content = message["content"]
|
102 |
+
history.append(f"{role}: {content}")
|
103 |
+
return "\n".join(history)
|
104 |
+
|
105 |
+
def display_course_card(course: Dict):
|
106 |
+
"""Display a single course recommendation in a card format."""
|
107 |
+
with st.container():
|
108 |
+
# Add a light background and padding
|
109 |
+
with st.container():
|
110 |
+
st.markdown("""
|
111 |
+
<style>
|
112 |
+
.course-card {
|
113 |
+
background-color: #f8f9fa;
|
114 |
+
padding: 1rem;
|
115 |
+
border-radius: 0.5rem;
|
116 |
+
margin-bottom: 1rem;
|
117 |
+
}
|
118 |
+
</style>
|
119 |
+
""", unsafe_allow_html=True)
|
120 |
+
|
121 |
+
with st.container():
|
122 |
+
st.markdown('<div class="course-card">', unsafe_allow_html=True)
|
123 |
+
|
124 |
+
# Course title
|
125 |
+
st.markdown(f"### {course['title']}")
|
126 |
+
|
127 |
+
col1, col2 = st.columns(2)
|
128 |
+
|
129 |
+
with col1:
|
130 |
+
# Handle categories - convert to list if string
|
131 |
+
categories = course.get('categories', 'N/A')
|
132 |
+
if isinstance(categories, str):
|
133 |
+
# Split by comma if it's a comma-separated string
|
134 |
+
categories = [cat.strip() for cat in categories.split(',')]
|
135 |
+
elif not isinstance(categories, list):
|
136 |
+
categories = [str(categories)]
|
137 |
+
|
138 |
+
# Display categories as bullet points if multiple
|
139 |
+
if len(categories) > 1:
|
140 |
+
st.markdown("**Categories:**")
|
141 |
+
for category in categories:
|
142 |
+
st.markdown(f"- {category}")
|
143 |
+
else:
|
144 |
+
st.markdown(f"**Category:** {categories[0]}")
|
145 |
+
|
146 |
+
st.markdown(f"**Lessons:** {course.get('lessons', 'N/A')}")
|
147 |
+
|
148 |
+
with col2:
|
149 |
+
st.markdown(f"**Price:** {course.get('price', 'N/A')}")
|
150 |
+
if 'url' in course:
|
151 |
+
st.markdown(f"**[Visit Course]({course['url']})**")
|
152 |
+
|
153 |
+
st.markdown('</div>', unsafe_allow_html=True)
|
154 |
+
|
155 |
+
st.markdown("---")
|
156 |
+
|
157 |
+
def main():
|
158 |
+
st.set_page_config(
|
159 |
+
page_title="Course Recommender",
|
160 |
+
page_icon="π",
|
161 |
+
layout="wide"
|
162 |
+
)
|
163 |
+
|
164 |
+
st.title("π AI Course Recommender")
|
165 |
+
|
166 |
+
# Initialize session state
|
167 |
+
initialize_session_state()
|
168 |
+
|
169 |
+
# Display collection info
|
170 |
+
collection = st.session_state.course_advisor.collection
|
171 |
+
st.sidebar.info(f"Connected to database with {collection.count()} courses")
|
172 |
+
|
173 |
+
# Sidebar
|
174 |
+
with st.sidebar:
|
175 |
+
st.header("Settings")
|
176 |
+
|
177 |
+
# API key input
|
178 |
+
api_key = st.text_input("Enter Groq API Key",
|
179 |
+
type="password",
|
180 |
+
value=st.session_state.api_key)
|
181 |
+
if api_key != st.session_state.api_key:
|
182 |
+
st.session_state.api_key = api_key
|
183 |
+
|
184 |
+
# Clear chat button
|
185 |
+
if st.button("Clear Chat History"):
|
186 |
+
st.session_state.messages = []
|
187 |
+
|
188 |
+
# Main chat interface
|
189 |
+
st.header("Chat with AI Course Advisor")
|
190 |
+
|
191 |
+
# Display chat history
|
192 |
+
for message in st.session_state.messages:
|
193 |
+
with st.chat_message(message["role"]):
|
194 |
+
st.markdown(message["content"])
|
195 |
+
|
196 |
+
# Chat input
|
197 |
+
if prompt := st.chat_input("What would you like to learn?"):
|
198 |
+
# Check if API key is provided
|
199 |
+
if not st.session_state.api_key:
|
200 |
+
st.error("Please enter your Groq API key in the sidebar.")
|
201 |
+
return
|
202 |
+
|
203 |
+
# Add user message to chat history
|
204 |
+
st.session_state.messages.append({"role": "user", "content": prompt})
|
205 |
+
|
206 |
+
with st.chat_message("user"):
|
207 |
+
st.markdown(prompt)
|
208 |
+
|
209 |
+
# Get AI response
|
210 |
+
with st.chat_message("assistant"):
|
211 |
+
with st.spinner("Thinking..."):
|
212 |
+
# Get formatted chat history
|
213 |
+
chat_history = get_chat_history()
|
214 |
+
|
215 |
+
# Query courses with chat history
|
216 |
+
response = st.session_state.course_advisor.query_courses(
|
217 |
+
prompt,
|
218 |
+
chat_history,
|
219 |
+
st.session_state.api_key
|
220 |
+
)
|
221 |
+
|
222 |
+
# Display AI recommendation
|
223 |
+
st.markdown(response['llm_response'])
|
224 |
+
|
225 |
+
# Display course cards if any courses were retrieved
|
226 |
+
if response['retrieved_courses']:
|
227 |
+
st.markdown("### π Recommended Courses")
|
228 |
+
for course in response['retrieved_courses']:
|
229 |
+
display_course_card(course)
|
230 |
+
|
231 |
+
# Add assistant response to chat history
|
232 |
+
st.session_state.messages.append({
|
233 |
+
"role": "assistant",
|
234 |
+
"content": response['llm_response'] + "\n\n" + "### Recommended Courses\n" +
|
235 |
+
"\n".join([f"- {course['title']}" for course in response['retrieved_courses']])
|
236 |
+
})
|
237 |
+
|
238 |
+
if __name__ == "__main__":
|
239 |
+
main()
|
chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/data_level0.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a13e72541800c513c73dccea69f79e39cf4baef4fa23f7e117c0d6b0f5f99670
|
3 |
+
size 3212000
|
chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/header.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0ec6df10978b056a10062ed99efeef2702fa4a1301fad702b53dd2517103c746
|
3 |
+
size 100
|
chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/length.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2823f194bebd05c68f5f4e78e39d91ac3064fb61f9163082a55db490fea91b56
|
3 |
+
size 4000
|
chroma_db/bca992f1-533f-4712-9825-dffc5cb4917e/link_lists.bin
ADDED
File without changes
|
chroma_db/chroma.sqlite3
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8973c5639f318e93a15f9963255d93350efdf75478f940cd8410ee3195646e07
|
3 |
+
size 2293760
|
requiremnts.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
streamlit
|
2 |
+
chromadb
|
3 |
+
groq
|