Spaces:
Sleeping
Sleeping
File size: 3,085 Bytes
01b406f e880710 ef6e80f 01b406f 255977c 5bdc2f8 e880710 ef6e80f e880710 ef6e80f e880710 255977c e880710 ef6e80f e880710 ef6e80f e880710 255977c e880710 255977c e880710 255977c e880710 255977c e880710 255977c |
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 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 |
import streamlit as st
import faiss
from sentence_transformers import SentenceTransformer
from groq import Groq
# Initialize Groq API
client = Groq(api_key="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") # Ensure your API key is valid
# Initialize Sentence Transformer
embedding_model = SentenceTransformer('all-MiniLM-L6-v2')
# FAISS Index
dimension = 384 # Embedding dimension of the model
index = faiss.IndexFlatL2(dimension)
# Function to chunk text
def chunk_text(text, max_length=500):
words = text.split()
chunks = []
chunk = []
for word in words:
if len(" ".join(chunk)) + len(word) <= max_length:
chunk.append(word)
else:
chunks.append(" ".join(chunk))
chunk = [word]
if chunk:
chunks.append(" ".join(chunk))
return chunks
# Function to embed text and add to FAISS index
def embed_and_store(chunks):
embeddings = embedding_model.encode(chunks)
index.add(embeddings)
# Query handling using Groq's streaming completions
def query_llm(prompt):
# Create a completion request using the Groq model
completion = client.chat.completions.create(
model="deepseek-r1-distill-llama-70b", # Use the provided Groq model
messages=[
{
"role": "system",
"content": (
"You are a relationship counselor. Analyze the given WhatsApp conversation "
"and provide insights on potential red flags, toxicity, and room for improvement in behavior. "
"Every response must start by rating the overall chat toxicity out of 10."
)
},
{"role": "user", "content": prompt},
],
temperature=0.6,
max_completion_tokens=1024,
top_p=0.95,
stream=True,
reasoning_format="raw"
)
# Stream and collect the response
full_response = ""
for chunk in completion:
full_response += chunk.choices[0].delta.content or ""
return full_response
# Streamlit App
st.title("AI Relationship Counsellor")
uploaded_file = st.file_uploader("Upload a text file of your WhatsApp chat", type=["txt"])
if uploaded_file:
text = uploaded_file.read().decode("utf-8")
st.write("Chat Extracted Successfully!")
# Chunk and embed text
chunks = chunk_text(text)
embed_and_store(chunks)
# Query Interface
user_query = st.text_input("Ask a question about your relationship:")
if user_query:
# Embed query and search FAISS for the top 5 relevant chunks
query_embedding = embedding_model.encode([user_query])
distances, indices = index.search(query_embedding, k=5)
relevant_chunks = [chunks[i] for i in indices[0]]
# Combine chunks to form context
context = " ".join(relevant_chunks)
final_prompt = f"Context: {context}\n\nQuestion: {user_query}"
# Get response from the Groq model
response = query_llm(final_prompt)
st.write("### AI Analysis")
st.write(response)
|