update

Dockerfile

@@ -0,0 +1,31 @@
+# Use Python 3.9 slim image
+FROM python:3.9-slim
+
+# Set working directory
+WORKDIR /app
+
+# Set environment variables
+ENV PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    PORT=8000
+
+# Install system dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy requirements first to leverage Docker cache
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy project files
+COPY . .
+
+# Create necessary directories
+RUN mkdir -p session_data session_summaries vector_db models
+
+# Expose the port
+EXPOSE 8000
+
+# Command to run the application
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8000"] 

README.md

@@ -1,257 +1,10 @@
-# Mental Health Support Chatbot
-
-A context-aware mental health support chatbot that provides therapeutic responses based on user emotions and maintains conversation history.
-
-## Features
-
-- Emotion detection using state-of-the-art NLP models
-- Context-aware responses
-- Conversation memory
-- Therapeutic techniques integration
-- Risk flag detection and crisis intervention
-  - Automatic detection of high-risk messages
-  - Immediate crisis response protocol
-  - Professional support referral system
-  - Emergency contact information
-- RESTful API interface
-- Session management and summaries
-- User reply tracking for another depression and anxiety detection from text.
-
-
-## Risk Flag Detection
-
-The chatbot automatically monitors messages for potential risk indicators and provides appropriate crisis intervention responses.
-
-### Risk Indicators
-The system detects various risk-related keywords and phrases, including but not limited to:
-- Self-harm references
-- Suicidal ideation
-- Extreme emotional distress
-- Crisis situations
-
-### Crisis Response Protocol
-When risk flags are detected:
-1. Immediate crisis response is triggered
-2. User is provided with:
-   - Emergency contact information
-   - Professional support options
-   - Immediate coping strategies
-3. Option to connect with licensed professionals
-4. Grounding exercises and calming techniques
-
-### Example Crisis Response
-```json
-{
-    "response":"I'm really sorry you're feeling this way — it sounds incredibly heavy,and I want you to know that you're not alone. You don't have to face this by yourself.Our app has licensed mental health professionals ready to support you.I can connect you with one right now if you'd like.Would you like to connect with a professional now,or would you rather keep talking with me for a bit? Either way, I'm here for you.",
-  "session_id": "user123_20240314103000",
-  "risk_detected": true,
-  "crisis_protocol_activated": true
-}
-```
-
-## Setup
-
-1. Install the required dependencies:
-```bash
-pip install -r requirements.txt
-```
-
-2. Download the required NLTK data:
-```bash
-python -m nltk.downloader punkt
-```
-
-3. Run the chatbot server:
-```bash
-python app.py
-```
-
-The server will start on `http://127.0.0.1:8000`
-
-## API Documentation
-
-### Base URL
-```
-http://127.0.0.1:8000
-```
-
-### API Endpoints
-
-#### 1. Start a Session
-```http
-POST /start_session?user_id={user_id}
-```
-
-Example:
-```bash
-curl -X 'POST' \
-  'http://127.0.0.1:8000/start_session?user_id=user123' \
-  -H 'accept: application/json'
-```
-
-Response:
-```json
-{
-    "response": "Hello! I'm here to support you today. How have you been feeling lately?",
-    "session_id": "user123_20240314103000"
-}
-```
-
-#### 2. Send a Message
-```http
-POST /send_message
-Content-Type: application/json
-
-{
-    "user_id": "user123",
-    "message": "I'm feeling anxious today"
-}
-```
-
-Example:
-```bash
-curl -X 'POST' \
-  'http://127.0.0.1:8000/send_message' \
-  -H 'accept: application/json' \
-  -H 'Content-Type: application/json' \
-  -d '{
-    "user_id": "user123",
-    "message": "I'\''m feeling anxious today"
-  }'
-```
-
-Response:
-```json
-{
-    "response": "I understand you're feeling anxious. Can you tell me more about what's causing this?",
-    "session_id": "user123_20240314103000"
-}
-```
-
-#### 3. Get User Replies
-```http
-GET /user_replies/{user_id}
-```
-
-Example:
-```bash
-curl -X 'GET' \
-  'http://127.0.0.1:8000/user_replies/user123' \
-  -H 'accept: application/json'
-```
-
-Response:
-```json
-{
-    "user_id": "user123",
-    "timestamp": "2024-03-14T10:30:00",
-    "replies": [
-        {
-            "text": "I'm feeling anxious today",
-            "timestamp": "2024-03-14T10:30:00",
-            "session_id": "user123_20240314103000"
-        }
-    ]
-}
-```
-
-#### 4. Get Session Summary
-```http
-GET /session_summary/{session_id}?include_summary={boolean}&include_recommendations={boolean}&include_emotions={boolean}&include_characteristics={boolean}&include_duration={boolean}&include_phase={boolean}
-```
-
-Example:
-```bash
-curl -X 'GET' \
-  'http://127.0.0.1:8000/session_summary/user123_20240314103000?include_summary=true&include_recommendations=true&include_emotions=true&include_characteristics=false&include_duration=false&include_phase=false' \
-  -H 'accept: application/json'
-```
-
-Response:
-```json
-{
-    "session_id": "user123_20240314103000",
-    "user_id": "user123",
-    "start_time": "2024-03-14T10:30:00",
-    "end_time": "2024-03-14T10:45:00",
-    "summary": "Session focused on anxiety management...",
-    "recommendations": [
-        "Practice deep breathing exercises",
-        "Consider journaling your thoughts"
-    ],
-    "primary_emotions": ["anxiety", "stress"],
-    "emotion_progression": ["anxiety", "calm"],
-    "duration_minutes": 0.0,
-    "current_phase": "unknown",
-    "session_characteristics": {}
-}
-```
-
-#### 5. End Session
-```http
-POST /end_session?user_id={user_id}
-```
-
-Example:
-```bash
-curl -X 'POST' \
-  'http://127.0.0.1:8000/end_session?user_id=user123' \
-  -H 'accept: application/json'
-```
-
-Response: Complete session summary with all fields.
-
-#### 6. Health Check
-```http
-GET /health
-```
-
-Example:
-```bash
-curl -X 'GET' \
-  'http://127.0.0.1:8000/health' \
-  -H 'accept: application/json'
-```
-
-Response:
-```json
-{
-    "status": "healthy"
-}
-```
-
-## Integration Guidelines
-
-### Best Practices
-1. Always store the `session_id` returned from `/start_session`
-2. Use the same `user_id` throughout a conversation
-3. Include appropriate error handling for API responses
-4. Monitor the health endpoint for system status
-
-### Error Handling
-The API returns standard HTTP status codes:
-- 200: Success
-- 400: Bad Request
-- 404: Not Found
-- 500: Internal Server Error
-
-Error responses include a detail message:
-```json
-{
-    "detail": "Error message here"
-}
-```
-
-
-## Important Notes
-
-- This is not a replacement for professional mental health care
-- Always seek professional help for serious mental health concerns
-
-
-## Privacy and Security
-
-- Conversations are stored in memory only
-- No personal data is permanently stored
-- The system is designed to be HIPAA-compliant
-- Users are identified by unique IDs only
+---
+title: Conversational Chat
+emoji: ⚡
+colorFrom: red
+colorTo: pink
+sdk: docker
+pinned: false
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

guidelines.txt

@@ -0,0 +1,107 @@
+Therapeutic Guidelines: 
+1. Build Trust and Rapport
+   
+   Begin with warmth and understanding.
+
+   Use active listening: reflect back emotions and key points.
+
+   Be supportive and non-threatening in tone.
+   
+   Always keep the tone calm, supportive, and emotionally intelligent.
+
+   Empower users to explore their own thoughts and solutions.
+
+   Ask open-ended questions to deepen self-reflection.
+
+   Avoid giving commands or rigid advice.
+   
+   Avoid assumptions based on culture, gender, or personal history.
+
+   Create psychological safety — reassure the user that their thoughts and emotions are welcome and valid.
+
+   
+2. Be Non-Judgmental
+
+   Accept all emotions and experiences without criticism.
+
+   Never blame or shame the user.
+
+   Normalize their feelings when appropriate
+   
+3. Use Evidence-Based Techniques
+   
+   Apply suitable techniques such as:
+     1. Cognitive Behavioral Therapy (CBT)
+         Help users identify negative thought patterns (cognitive distortions) and reframe them:
+
+         “Let’s try to challenge that thought — is there evidence that supports or contradicts it?”
+
+         “What might be a more balanced way to look at this?”
+
+     2. Dialectical Behavior Therapy (DBT)
+         Focus on emotional regulation, distress tolerance, and mindfulness:
+
+         “Let’s take a moment to breathe and notice what you’re feeling without judgment.”
+
+         “What can you do right now to self-soothe or ground yourself?”
+
+     3. Acceptance and Commitment Therapy (ACT)
+         Promote acceptance of thoughts and values-based living:
+
+         “Instead of fighting that thought, can we observe it and let it be?”
+
+         “What matters to you right now? What small step can you take in that direction?”
+
+     4. Motivational Interviewing
+         Help ambivalent users explore change:
+
+         “On a scale from 1 to 10, how ready do you feel to make a change?”
+
+         “What would it take to move one step closer?”
+
+
+4. Structured Conversation Flow
+   Begin with empathy → explore the problem → validate emotions → apply a therapeutic tool → summarize insight or coping step.
+
+   End each message with a question or reflection prompt to continue engagement.
+
+
+5. Add Actionable Suggestions
+
+     Offer gentle, realistic, and practical steps the user can try.
+
+     Tailor suggestions to their emotional state — prioritize simplicity and emotional safety.
+
+     Use empowering language that invites, not instructs:
+
+         “Would you be open to trying…?”
+
+         “Some people find this helpful — would you like to explore it together?”
+
+     Examples of actionable suggestions include:
+
+         Grounding Techniques
+             “Can you name five things you see around you right now, four things you can touch, three you can hear, two you can smell, and one you can taste?”
+
+         Mindful Breathing
+             “Let’s try a simple breathing exercise: inhale slowly for 4 counts, hold for 4, exhale for 4. Can we do this together for a few rounds?”
+
+     Journaling Prompts
+         “Would writing down your thoughts help make sense of what you're feeling? You might start with: ‘Right now, I’m feeling… because…’”
+
+     Self-Compassion Reminders
+         “Can you speak to yourself the way you would to a friend going through this?”
+
+     Behavioral Activation
+         “Sometimes doing one small activity, even if it feels meaningless at first, can help shift your energy. What’s one thing you could do today that used to bring you comfort?”
+
+     Connection Check-In
+         “Is there someone you trust that you might feel comfortable talking to or spending time with today, even briefly?”
+
+     End with an open tone:
+
+         “How does that sound to you?”
+
+         “Would you like to try that and let me know how it goes?”
+
+   

mental_health_chatbot.log

@@ -0,0 +1,782 @@
+2025-04-16 20:40:51,091 - __main__ - INFO - Using device: cuda
+2025-04-16 20:40:51,091 - __main__ - INFO - Loading emotion detection model
+2025-04-16 20:40:51,872 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:40:52,900 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-16 20:40:54,064 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:41:04,152 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-16 20:41:04,455 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-16 20:41:05,633 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-16 20:41:08,333 - __main__ - INFO - Setting up FAISS vector database
+2025-04-16 20:41:08,663 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-16 20:41:08,728 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-16 20:41:08,741 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-16 20:41:08,746 - __main__ - WARNING - No guidelines file provided, using empty vector store
+2025-04-16 20:49:53,663 - __main__ - INFO - Using device: cuda
+2025-04-16 20:49:53,663 - __main__ - INFO - Loading emotion detection model
+2025-04-16 20:49:54,306 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:49:55,317 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-16 20:49:56,722 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:50:05,931 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-16 20:50:06,203 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-16 20:50:07,402 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-16 20:50:10,384 - __main__ - INFO - Setting up FAISS vector database
+2025-04-16 20:50:10,385 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-16 20:50:10,445 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-16 20:50:10,458 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-16 20:50:10,461 - __main__ - INFO - Loaded existing vector database
+2025-04-16 20:53:57,905 - __main__ - INFO - Using device: cuda
+2025-04-16 20:53:57,905 - __main__ - INFO - Loading emotion detection model
+2025-04-16 20:53:58,645 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:53:59,640 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-16 20:54:00,686 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:54:10,841 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-16 20:54:11,142 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-16 20:54:12,244 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-16 20:54:15,613 - __main__ - INFO - Setting up FAISS vector database
+2025-04-16 20:54:15,619 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-16 20:54:15,670 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-16 20:54:15,678 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-16 20:54:15,680 - __main__ - INFO - Loaded existing vector database
+2025-04-16 20:56:31,196 - __main__ - INFO - Using device: cuda
+2025-04-16 20:56:31,196 - __main__ - INFO - Loading emotion detection model
+2025-04-16 20:56:32,364 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:56:33,303 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-16 20:56:34,880 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-16 20:56:44,016 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-16 20:56:44,374 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-16 20:56:45,451 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-16 20:56:48,249 - __main__ - INFO - Setting up FAISS vector database
+2025-04-16 20:56:48,252 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-16 20:56:48,274 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-16 20:56:48,282 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-16 20:56:48,284 - __main__ - INFO - Loaded existing vector database
+2025-04-16 20:56:48,322 - __main__ - INFO - Session started for user cli_user_20250416205648
+2025-04-18 16:02:11,023 - __main__ - INFO - Using device: cuda
+2025-04-18 16:02:11,023 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:02:12,079 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:02:13,129 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:02:14,361 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:02:24,172 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:02:24,514 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:02:25,616 - __main__ - INFO - Loading summary model
+2025-04-18 16:22:26,761 - __main__ - INFO - Initializing FlowManager
+2025-04-18 16:22:26,762 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 16:22:26,764 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 16:22:30,903 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 16:22:30,914 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 16:22:31,039 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 16:22:31,045 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 16:22:31,074 - __main__ - INFO - Loaded existing vector database
+2025-04-18 16:22:31,087 - conversation_flow - INFO - Initialized new session for user cli_user_20250418162231
+2025-04-18 16:22:31,087 - __main__ - INFO - Session started for user cli_user_20250418162231
+2025-04-18 16:28:53,111 - __main__ - INFO - Using device: cuda
+2025-04-18 16:28:53,111 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:29:03,485 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:29:04,516 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:29:05,512 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:29:14,677 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:29:14,987 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:29:16,117 - __main__ - INFO - Loading summary model
+2025-04-18 16:31:42,623 - __main__ - INFO - Using device: cuda
+2025-04-18 16:31:42,630 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:31:43,302 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:31:44,315 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:31:45,437 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:31:54,477 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:31:54,744 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:31:55,750 - __main__ - INFO - Loading summary model
+2025-04-18 16:33:51,319 - __main__ - INFO - Using device: cuda
+2025-04-18 16:33:51,320 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:33:52,044 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:33:53,063 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:33:54,159 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:34:03,223 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:34:03,556 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:34:04,651 - __main__ - INFO - Loading summary model
+2025-04-18 16:34:05,893 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:39:59,658 - __main__ - INFO - Using device: cuda
+2025-04-18 16:39:59,659 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:40:00,514 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:40:01,521 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:40:03,059 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:40:12,212 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:40:12,491 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:40:13,567 - __main__ - INFO - Loading summary model
+2025-04-18 16:40:16,727 - __main__ - INFO - Initializing FlowManager
+2025-04-18 16:40:16,727 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 16:43:27,852 - __main__ - INFO - Using device: cuda
+2025-04-18 16:43:27,855 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:43:28,440 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:43:29,386 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:43:30,348 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:43:39,286 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:43:39,558 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:43:40,570 - __main__ - INFO - Loading summary model
+2025-04-18 16:43:43,510 - __main__ - INFO - Initializing FlowManager
+2025-04-18 16:43:43,518 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 16:43:43,520 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 16:43:46,271 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 16:43:46,276 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 16:43:46,343 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 16:43:46,351 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 16:43:46,355 - __main__ - INFO - Loaded existing vector database
+2025-04-18 16:43:46,356 - conversation_flow - INFO - Initialized new session for user cli_user_20250418164346
+2025-04-18 16:43:46,357 - __main__ - INFO - Session started for user cli_user_20250418164346
+2025-04-18 16:48:37,587 - __main__ - INFO - Using device: cuda
+2025-04-18 16:48:37,587 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:48:38,210 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:48:39,162 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:48:40,193 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:48:49,130 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:48:49,437 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:48:50,554 - __main__ - INFO - Loading summary model
+2025-04-18 16:48:53,718 - __main__ - INFO - Initializing FlowManager
+2025-04-18 16:48:53,718 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 16:48:53,718 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 16:49:00,071 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 16:49:00,074 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 16:49:00,130 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 16:49:00,141 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 16:49:00,144 - __main__ - INFO - Loaded existing vector database
+2025-04-18 16:49:00,145 - conversation_flow - INFO - Initialized new session for user cli_user_20250418164900
+2025-04-18 16:49:00,145 - __main__ - INFO - Session started for user cli_user_20250418164900
+2025-04-18 16:52:02,476 - __main__ - INFO - Using device: cuda
+2025-04-18 16:52:02,476 - __main__ - INFO - Loading emotion detection model
+2025-04-18 16:52:03,111 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:52:04,143 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 16:52:05,213 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 16:52:14,106 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 16:52:14,438 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 16:52:15,455 - __main__ - INFO - Loading summary model
+2025-04-18 16:52:18,449 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 16:52:18,449 - __main__ - INFO - Initializing FlowManager
+2025-04-18 16:52:18,449 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 16:52:18,454 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 16:52:21,626 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 16:52:21,637 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 16:52:21,678 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 16:52:21,699 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 16:52:21,702 - __main__ - INFO - Loaded existing vector database
+2025-04-18 16:52:21,703 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 16:52:21,704 - conversation_flow - INFO - Initialized new session for user cli_user_20250418165221
+2025-04-18 16:52:21,704 - __main__ - INFO - Session started for user cli_user_20250418165221
+2025-04-18 17:18:39,952 - __main__ - INFO - Using device: cuda
+2025-04-18 17:18:39,952 - __main__ - INFO - Loading emotion detection model
+2025-04-18 17:18:40,598 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 17:18:41,654 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 17:18:42,682 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 17:18:51,948 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 17:18:52,282 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 17:18:53,411 - __main__ - INFO - Loading summary model
+2025-04-18 17:18:56,632 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 17:18:56,632 - __main__ - INFO - Initializing FlowManager
+2025-04-18 17:18:56,632 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 17:18:56,632 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 17:19:00,694 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 17:19:00,698 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 17:19:00,749 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 17:19:00,760 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 17:19:00,763 - __main__ - INFO - Loaded existing vector database
+2025-04-18 17:19:00,764 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 17:19:00,765 - conversation_flow - INFO - Initialized new session for user cli_user_20250418171900
+2025-04-18 17:19:00,765 - __main__ - INFO - Session started for user cli_user_20250418171900
+2025-04-18 20:42:57,848 - __main__ - INFO - Using device: cuda
+2025-04-18 20:42:57,848 - __main__ - INFO - Loading emotion detection model
+2025-04-18 20:43:02,595 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 20:43:03,524 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 20:43:04,598 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 20:43:12,915 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 20:43:13,129 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 20:43:14,236 - __main__ - INFO - Loading summary model
+2025-04-18 20:43:17,220 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 20:43:17,220 - __main__ - INFO - Initializing FlowManager
+2025-04-18 20:43:17,220 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 20:43:17,233 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 20:43:21,870 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 20:43:21,870 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 20:43:21,929 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 20:43:21,944 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 20:43:21,953 - __main__ - INFO - Loaded existing vector database
+2025-04-18 20:43:21,954 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 20:43:21,955 - conversation_flow - INFO - Initialized new session for user cli_user_20250418204321
+2025-04-18 20:43:21,955 - __main__ - INFO - Session started for user cli_user_20250418204321
+2025-04-18 20:44:10,846 - conversation_flow - ERROR - Error detecting topics with LLM: 'HuggingFacePipeline' object has no attribute 'get_llm_response'
+2025-04-18 20:44:59,396 - conversation_flow - ERROR - Error detecting topics with LLM: 'HuggingFacePipeline' object has no attribute 'get_llm_response'
+2025-04-18 20:45:25,345 - conversation_flow - ERROR - Error detecting topics with LLM: 'HuggingFacePipeline' object has no attribute 'get_llm_response'
+2025-04-18 20:45:47,579 - conversation_flow - ERROR - Error detecting topics with LLM: 'HuggingFacePipeline' object has no attribute 'get_llm_response'
+2025-04-18 20:46:06,205 - conversation_flow - ERROR - Error detecting topics with LLM: 'HuggingFacePipeline' object has no attribute 'get_llm_response'
+2025-04-18 21:01:48,815 - __main__ - INFO - Using device: cuda
+2025-04-18 21:01:48,817 - __main__ - INFO - Loading emotion detection model
+2025-04-18 21:01:50,288 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:01:51,205 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 21:01:52,274 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:02:00,508 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 21:02:00,733 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 21:02:01,861 - __main__ - INFO - Loading summary model
+2025-04-18 21:02:04,829 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 21:02:04,829 - __main__ - INFO - Initializing FlowManager
+2025-04-18 21:02:04,829 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 21:02:04,829 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 21:02:07,509 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 21:02:07,513 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 21:02:07,571 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 21:02:07,576 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 21:02:07,580 - __main__ - INFO - Loaded existing vector database
+2025-04-18 21:02:07,581 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 21:02:07,582 - conversation_flow - INFO - Initialized new session for user cli_user_20250418210207
+2025-04-18 21:02:07,582 - __main__ - INFO - Session started for user cli_user_20250418210207
+2025-04-18 21:09:03,887 - __main__ - INFO - Using device: cuda
+2025-04-18 21:09:03,887 - __main__ - INFO - Loading emotion detection model
+2025-04-18 21:09:05,546 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:09:06,525 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 21:09:07,498 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:09:15,645 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 21:09:15,852 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 21:09:16,802 - __main__ - INFO - Loading summary model
+2025-04-18 21:09:19,599 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 21:09:19,599 - __main__ - INFO - Initializing FlowManager
+2025-04-18 21:09:19,599 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 21:09:19,605 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 21:09:32,385 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 21:09:32,401 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 21:09:32,443 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 21:09:32,458 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 21:09:32,465 - __main__ - INFO - Loaded existing vector database
+2025-04-18 21:09:32,465 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 21:09:32,465 - conversation_flow - INFO - Initialized new session for user cli_user_20250418210932
+2025-04-18 21:09:32,465 - __main__ - INFO - Session started for user cli_user_20250418210932
+2025-04-18 21:10:12,360 - __main__ - ERROR - Failed to generate session summary: Object of type ConversationPhase is not JSON serializable
+2025-04-18 21:19:08,728 - __main__ - INFO - Using device: cuda
+2025-04-18 21:19:08,728 - __main__ - INFO - Loading emotion detection model
+2025-04-18 21:19:09,386 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:19:10,380 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 21:19:11,771 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:19:20,833 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 21:19:21,122 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 21:19:22,118 - __main__ - INFO - Loading summary model
+2025-04-18 21:19:25,280 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 21:19:25,280 - __main__ - INFO - Initializing FlowManager
+2025-04-18 21:19:25,280 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 21:19:25,294 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 21:19:28,905 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 21:19:28,908 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 21:19:28,964 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 21:19:28,980 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 21:19:28,984 - __main__ - INFO - Loaded existing vector database
+2025-04-18 21:19:28,985 - __main__ - WARNING - Failed to load summary from .json: Expecting value: line 7 column 20 (char 147)
+2025-04-18 21:19:28,985 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 21:19:28,986 - conversation_flow - INFO - Initialized new session for user cli_user_20250418211928
+2025-04-18 21:19:28,986 - __main__ - INFO - Session started for user cli_user_20250418211928
+2025-04-18 21:26:19,114 - __main__ - INFO - Using device: cuda
+2025-04-18 21:26:19,114 - __main__ - INFO - Loading emotion detection model
+2025-04-18 21:26:19,762 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:26:20,784 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 21:26:21,847 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:26:30,681 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 21:26:31,011 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 21:26:31,996 - __main__ - INFO - Loading summary model
+2025-04-18 21:26:34,971 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 21:26:34,971 - __main__ - INFO - Initializing FlowManager
+2025-04-18 21:26:34,971 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 21:26:34,985 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 21:26:45,007 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 21:26:45,010 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 21:26:45,068 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 21:26:45,077 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 21:26:45,080 - __main__ - INFO - Loaded existing vector database
+2025-04-18 21:26:45,081 - __main__ - WARNING - Failed to load summary from .json: Expecting value: line 7 column 20 (char 147)
+2025-04-18 21:26:45,082 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 21:26:45,082 - conversation_flow - INFO - Initialized new session for user cli_user_20250418212645
+2025-04-18 21:26:45,082 - __main__ - INFO - Session started for user cli_user_20250418212645
+2025-04-18 21:32:34,109 - conversation_flow - INFO - User cli_user_20250418212645 transitioned from introduction to exploration: Time-based transition
+2025-04-18 21:58:42,487 - __main__ - INFO - Using device: cuda
+2025-04-18 21:58:42,492 - __main__ - INFO - Loading emotion detection model
+2025-04-18 21:58:43,126 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 21:58:44,158 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 21:58:45,213 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:08:32,721 - __main__ - INFO - Using device: cuda
+2025-04-18 22:08:32,721 - __main__ - INFO - Loading emotion detection model
+2025-04-18 22:08:38,582 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:08:39,309 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 22:08:42,392 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:08:47,815 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 22:08:48,105 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 22:08:49,156 - __main__ - INFO - Loading summary model
+2025-04-18 22:08:57,299 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 22:08:57,299 - __main__ - INFO - Initializing FlowManager
+2025-04-18 22:08:57,299 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 22:08:57,302 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 22:09:17,127 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 22:09:17,130 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 22:09:17,203 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 22:09:17,213 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 22:09:17,218 - __main__ - INFO - Loaded existing vector database
+2025-04-18 22:09:17,219 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 22:09:17,220 - conversation_flow - INFO - Initialized new session for user cli_user_20250418220917
+2025-04-18 22:09:17,220 - __main__ - INFO - Session started for user cli_user_20250418220917
+2025-04-18 22:17:05,900 - __main__ - INFO - Using device: cuda
+2025-04-18 22:17:05,900 - __main__ - INFO - Loading emotion detection model
+2025-04-18 22:17:06,561 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:17:07,562 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 22:17:08,643 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:17:17,695 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 22:17:18,024 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 22:17:19,055 - __main__ - INFO - Loading summary model
+2025-04-18 22:17:22,232 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 22:17:22,232 - __main__ - INFO - Initializing FlowManager
+2025-04-18 22:17:22,232 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 22:17:22,242 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 22:17:37,477 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 22:17:37,481 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 22:17:37,543 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 22:17:37,550 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 22:17:37,553 - __main__ - INFO - Loaded existing vector database
+2025-04-18 22:17:37,554 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 22:17:37,555 - conversation_flow - INFO - Initialized new session for user cli_user_20250418221737
+2025-04-18 22:17:37,555 - __main__ - INFO - Session started for user cli_user_20250418221737
+2025-04-18 22:18:57,039 - __main__ - INFO - Using device: cuda
+2025-04-18 22:18:57,040 - __main__ - INFO - Loading emotion detection model
+2025-04-18 22:18:59,206 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:19:00,202 - __main__ - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-18 22:19:01,317 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-18 22:19:10,383 - __main__ - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-18 22:19:10,680 - __main__ - INFO - Successfully loaded PEFT model
+2025-04-18 22:19:11,731 - __main__ - INFO - Loading summary model
+2025-04-18 22:19:20,329 - __main__ - INFO - Summary model loaded successfully
+2025-04-18 22:19:20,329 - __main__ - INFO - Initializing FlowManager
+2025-04-18 22:19:20,329 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-18 22:19:20,343 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-18 22:19:23,597 - __main__ - INFO - Setting up FAISS vector database
+2025-04-18 22:19:23,599 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-18 22:19:23,655 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-18 22:19:23,661 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-18 22:19:23,686 - __main__ - INFO - Loaded existing vector database
+2025-04-18 22:19:23,687 - __main__ - INFO - All models and components initialized successfully
+2025-04-18 22:19:23,688 - conversation_flow - INFO - Initialized new session for user cli_user_20250418221923
+2025-04-18 22:19:23,688 - __main__ - INFO - Session started for user cli_user_20250418221923
+2025-04-18 22:22:44,393 - conversation_flow - WARNING - Failed to parse session characteristics from LLM
+2025-04-18 22:32:18,080 - __main__ - ERROR - Failed to generate session summary: 'str' object has no attribute 'items'
+2025-04-19 20:34:55,476 - claude - INFO - Using device: cpu
+2025-04-19 20:34:55,476 - claude - INFO - Loading emotion detection model
+2025-04-19 20:34:57,058 - claude - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 20:35:34,008 - claude - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-19 20:35:34,086 - bitsandbytes.cextension - WARNING - The installed version of bitsandbytes was compiled without GPU support. 8-bit optimizers, 8-bit multiplication, and GPU quantization are unavailable.
+2025-04-19 20:35:34,458 - claude - INFO - Successfully loaded PEFT model
+2025-04-19 20:35:37,385 - claude - INFO - Loading summary model
+2025-04-19 20:35:38,798 - claude - INFO - Summary model loaded successfully
+2025-04-19 20:35:38,799 - claude - INFO - Initializing FlowManager
+2025-04-19 20:35:38,799 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 20:35:38,810 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-19 20:35:41,743 - claude - INFO - Setting up FAISS vector database
+2025-04-19 20:35:41,750 - faiss.loader - INFO - Loading faiss.
+2025-04-19 20:35:42,519 - faiss.loader - INFO - Successfully loaded faiss.
+2025-04-19 20:35:42,533 - claude - INFO - Loaded existing vector database
+2025-04-19 20:35:42,535 - claude - INFO - All models and components initialized successfully
+2025-04-19 20:37:49,972 - claude - INFO - Using device: cuda
+2025-04-19 20:37:49,973 - claude - INFO - Loading emotion detection model
+2025-04-19 20:37:50,809 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 20:37:51,983 - claude - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 20:37:53,346 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 20:38:03,080 - claude - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-19 20:38:03,408 - claude - INFO - Successfully loaded PEFT model
+2025-04-19 20:38:04,549 - claude - INFO - Loading summary model
+2025-04-19 20:38:07,765 - claude - INFO - Summary model loaded successfully
+2025-04-19 20:38:07,765 - claude - INFO - Initializing FlowManager
+2025-04-19 20:38:07,766 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 20:38:07,772 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-19 20:38:10,738 - claude - INFO - Setting up FAISS vector database
+2025-04-19 20:38:10,742 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-19 20:38:10,812 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-19 20:38:10,822 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-19 20:38:10,825 - claude - INFO - Loaded existing vector database
+2025-04-19 20:38:10,827 - claude - INFO - All models and components initialized successfully
+2025-04-19 20:40:27,294 - claude - INFO - Using device: cuda
+2025-04-19 20:40:27,295 - claude - INFO - Loading emotion detection model
+2025-04-19 20:40:27,946 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 20:40:28,924 - claude - INFO - Loading LLM model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 20:40:30,351 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 20:40:39,190 - claude - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-19 20:40:39,529 - claude - INFO - Successfully loaded PEFT model
+2025-04-19 20:40:40,687 - claude - INFO - Loading summary model
+2025-04-19 20:40:43,582 - claude - INFO - Summary model loaded successfully
+2025-04-19 20:40:43,583 - claude - INFO - Initializing FlowManager
+2025-04-19 20:40:43,584 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 20:40:43,589 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-19 20:40:47,142 - claude - INFO - Setting up FAISS vector database
+2025-04-19 20:40:47,146 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-19 20:40:47,206 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-19 20:40:47,214 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-19 20:40:47,218 - claude - INFO - Loaded existing vector database
+2025-04-19 20:40:47,219 - claude - INFO - All models and components initialized successfully
+2025-04-19 20:41:52,584 - conversation_flow - INFO - Initialized new session for user user_1
+2025-04-19 20:41:52,585 - claude - INFO - Session started for user user_1
+2025-04-19 20:44:44,549 - conversation_flow - INFO - Initialized new session for user test_user_20250419204444
+2025-04-19 20:44:44,550 - claude - INFO - Session started for user test_user_20250419204444
+2025-04-19 20:51:44,998 - conversation_flow - INFO - Initialized new session for user user_1
+2025-04-19 20:51:44,998 - claude - INFO - Session started for user user_1
+2025-04-19 21:22:26,351 - chatbot - INFO - Using device: cuda
+2025-04-19 21:22:26,352 - chatbot - INFO - Loading emotion detection model
+2025-04-19 21:22:27,213 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 21:22:28,233 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 21:22:29,310 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 21:22:38,570 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-19 21:22:38,810 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-19 21:22:39,906 - chatbot - INFO - Loading summary model
+2025-04-19 21:22:43,171 - chatbot - INFO - Summary model loaded successfully
+2025-04-19 21:22:43,171 - chatbot - INFO - Initializing FlowManager
+2025-04-19 21:22:43,172 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 21:22:43,177 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-19 21:22:46,003 - chatbot - INFO - Setting up FAISS vector database
+2025-04-19 21:22:46,007 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-19 21:22:46,075 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-19 21:22:46,082 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-19 21:22:46,085 - chatbot - INFO - Loaded existing vector database
+2025-04-19 21:22:46,086 - chatbot - INFO - All models and components initialized successfully
+2025-04-19 21:24:58,360 - chatbot - INFO - Using device: cuda
+2025-04-19 21:24:58,360 - chatbot - INFO - Loading emotion detection model
+2025-04-19 21:24:59,401 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 21:25:00,460 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 21:25:01,753 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 21:25:11,213 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-19 21:25:11,506 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-19 21:25:12,577 - chatbot - INFO - Loading summary model
+2025-04-19 21:25:15,561 - chatbot - INFO - Summary model loaded successfully
+2025-04-19 21:25:15,561 - chatbot - INFO - Initializing FlowManager
+2025-04-19 21:25:15,562 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 21:25:15,568 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-19 21:25:18,810 - chatbot - INFO - Setting up FAISS vector database
+2025-04-19 21:25:18,814 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-19 21:25:18,875 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-19 21:25:18,882 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-19 21:25:18,886 - chatbot - INFO - Loaded existing vector database
+2025-04-19 21:25:18,887 - chatbot - INFO - All models and components initialized successfully
+2025-04-19 21:25:45,461 - conversation_flow - INFO - Initialized new session for user test_user_20250419212545
+2025-04-19 21:25:45,462 - chatbot - INFO - Session started for user test_user_20250419212545
+2025-04-19 21:26:52,439 - conversation_flow - INFO - Initialized new session for user user_1
+2025-04-19 21:26:52,439 - chatbot - INFO - Session started for user user_1
+2025-04-19 23:03:08,804 - chatbot - INFO - Using device: cuda
+2025-04-19 23:03:08,805 - chatbot - INFO - Loading emotion detection model
+2025-04-19 23:03:09,619 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 23:03:10,663 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 23:03:11,775 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 23:03:21,298 - chatbot - INFO - Loading PEFT model from Hugging Face
+2025-04-19 23:03:34,745 - chatbot - INFO - Successfully loaded PEFT model from Hugging Face
+2025-04-19 23:03:35,902 - chatbot - INFO - Loading summary model
+2025-04-19 23:03:39,040 - chatbot - INFO - Summary model loaded successfully
+2025-04-19 23:03:39,040 - chatbot - INFO - Initializing FlowManager
+2025-04-19 23:03:39,041 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 23:03:39,042 - chatbot - INFO - Setting up FAISS vector database
+2025-04-19 23:05:16,618 - chatbot - INFO - Using device: cuda
+2025-04-19 23:05:16,618 - chatbot - INFO - Initializing embeddings
+2025-04-19 23:05:16,623 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-mpnet-base-v2
+2025-04-19 23:19:58,437 - chatbot - INFO - Using device: cuda
+2025-04-19 23:19:58,437 - chatbot - INFO - Loading emotion detection model
+2025-04-19 23:19:59,596 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 23:20:00,656 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 23:20:03,317 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 23:20:13,381 - chatbot - INFO - Loading PEFT model from Hugging Face
+2025-04-19 23:20:14,931 - chatbot - INFO - Successfully loaded PEFT model from Hugging Face
+2025-04-19 23:20:16,129 - chatbot - INFO - Loading summary model
+2025-04-19 23:20:20,632 - chatbot - INFO - Summary model loaded successfully
+2025-04-19 23:20:20,633 - chatbot - INFO - Initializing FlowManager
+2025-04-19 23:20:20,633 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 23:20:20,634 - chatbot - INFO - Setting up FAISS vector database
+2025-04-19 23:20:20,635 - chatbot - INFO - Initializing embeddings
+2025-04-19 23:20:20,639 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-mpnet-base-v2
+2025-04-19 23:25:29,794 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-19 23:25:29,868 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-19 23:25:29,877 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-19 23:25:29,881 - chatbot - INFO - Loaded existing vector database
+2025-04-19 23:25:29,883 - chatbot - INFO - All models and components initialized successfully
+2025-04-19 23:27:45,651 - conversation_flow - INFO - Initialized new session for user test_user_20250419232745
+2025-04-19 23:27:45,652 - chatbot - INFO - Session started for user test_user_20250419232745
+2025-04-19 23:27:46,102 - chatbot - ERROR - Error retrieving guidelines: 
+2025-04-19 23:27:46,143 - chatbot - ERROR - Error retrieving context: 
+2025-04-19 23:29:51,225 - conversation_flow - INFO - Initialized new session for user test_user_20250419232951
+2025-04-19 23:29:51,226 - chatbot - INFO - Session started for user test_user_20250419232951
+2025-04-19 23:29:51,303 - chatbot - ERROR - Error retrieving guidelines: 
+2025-04-19 23:29:51,342 - chatbot - ERROR - Error retrieving context: 
+2025-04-19 23:31:21,388 - chatbot - INFO - Using device: cuda
+2025-04-19 23:31:21,389 - chatbot - INFO - Loading emotion detection model
+2025-04-19 23:31:22,154 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 23:31:23,228 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-19 23:31:24,418 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-19 23:31:33,745 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-19 23:31:34,056 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-19 23:31:35,096 - chatbot - INFO - Loading summary model
+2025-04-19 23:31:38,449 - chatbot - INFO - Summary model loaded successfully
+2025-04-19 23:31:38,449 - chatbot - INFO - Initializing FlowManager
+2025-04-19 23:31:38,449 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-19 23:31:38,454 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-19 23:31:40,970 - chatbot - INFO - Setting up FAISS vector database
+2025-04-19 23:31:40,975 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-19 23:31:41,041 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-19 23:31:41,049 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-19 23:31:41,053 - chatbot - INFO - Loaded existing vector database
+2025-04-19 23:31:41,054 - chatbot - INFO - All models and components initialized successfully
+2025-04-19 23:31:58,013 - conversation_flow - INFO - Initialized new session for user test_user_20250419233158
+2025-04-19 23:31:58,014 - chatbot - INFO - Session started for user test_user_20250419233158
+2025-04-20 16:45:15,627 - chatbot - INFO - Using device: cuda
+2025-04-20 16:45:15,628 - chatbot - INFO - Loading emotion detection model
+2025-04-20 16:45:16,472 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 16:45:17,756 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-20 16:45:19,008 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 16:45:29,740 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-20 16:45:30,062 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-20 16:45:31,256 - chatbot - INFO - Loading summary model
+2025-04-20 16:45:34,713 - chatbot - INFO - Summary model loaded successfully
+2025-04-20 16:45:34,714 - chatbot - INFO - Initializing FlowManager
+2025-04-20 16:45:34,714 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-20 16:45:34,719 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-20 16:45:38,475 - chatbot - INFO - Setting up FAISS vector database
+2025-04-20 16:45:38,480 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-20 16:45:38,549 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-20 16:45:38,557 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-20 16:45:38,577 - chatbot - INFO - Loaded existing vector database
+2025-04-20 16:45:38,616 - chatbot - INFO - All models and components initialized successfully
+2025-04-20 16:45:51,568 - conversation_flow - INFO - Initialized new session for user test_user_20250420164551
+2025-04-20 16:45:51,569 - chatbot - INFO - Session started for user test_user_20250420164551
+2025-04-20 17:02:33,896 - chatbot - INFO - Using device: cuda
+2025-04-20 17:02:33,896 - chatbot - INFO - Loading emotion detection model
+2025-04-20 17:02:34,494 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:02:35,523 - chatbot - INFO - Loading LLAMA model: nada013/mental-health-chatbot
+2025-04-20 17:06:34,211 - chatbot - INFO - Using device: cuda
+2025-04-20 17:06:34,211 - chatbot - INFO - Loading emotion detection model
+2025-04-20 17:06:35,028 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:06:35,678 - chatbot - INFO - Loading LLAMA model: nada013/mental-health-chatbot
+2025-04-20 17:08:08,185 - chatbot - INFO - Using device: cuda
+2025-04-20 17:08:08,185 - chatbot - INFO - Loading emotion detection model
+2025-04-20 17:08:08,734 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:08:09,362 - chatbot - INFO - Loading LLAMA model: nada013/mental-health-chatbot
+2025-04-20 17:10:44,643 - chatbot - INFO - Using device: cuda
+2025-04-20 17:10:44,644 - chatbot - INFO - Loading emotion detection model
+2025-04-20 17:10:45,420 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:10:46,075 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-20 17:10:47,089 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:10:56,278 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-20 17:10:56,578 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-20 17:10:57,961 - chatbot - INFO - Loading summary model
+2025-04-20 17:11:01,854 - chatbot - INFO - Summary model loaded successfully
+2025-04-20 17:11:01,855 - chatbot - INFO - Initializing FlowManager
+2025-04-20 17:11:01,855 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-20 17:11:01,861 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-20 17:15:58,375 - chatbot - INFO - Using device: cuda
+2025-04-20 17:15:58,375 - chatbot - INFO - Loading emotion detection model
+2025-04-20 17:15:59,033 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:16:00,047 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-20 17:16:00,049 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-20 17:16:00,443 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-20 17:16:09,546 - chatbot - INFO - Loading tokenizer
+2025-04-20 17:16:10,189 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-20 17:16:10,492 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-20 17:16:10,495 - chatbot - INFO - Loading summary model
+2025-04-20 17:16:13,817 - chatbot - INFO - Summary model loaded successfully
+2025-04-20 17:16:13,817 - chatbot - INFO - Initializing FlowManager
+2025-04-20 17:16:13,817 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-20 17:16:13,824 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-20 17:16:16,265 - chatbot - INFO - Setting up FAISS vector database
+2025-04-20 17:16:16,269 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-20 17:16:16,336 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-20 17:16:16,344 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-20 17:16:16,348 - chatbot - INFO - Loaded existing vector database
+2025-04-20 17:16:16,350 - chatbot - INFO - All models and components initialized successfully
+2025-04-20 17:16:30,886 - conversation_flow - INFO - Initialized new session for user test_user_20250420171630
+2025-04-20 17:16:30,886 - chatbot - INFO - Session started for user test_user_20250420171630
+2025-04-28 13:34:59,994 - chatbot - INFO - Using device: cuda
+2025-04-28 13:34:59,994 - chatbot - INFO - Loading emotion detection model
+2025-04-28 13:35:00,828 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-28 13:35:01,872 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-28 13:35:01,872 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-28 13:35:02,679 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-28 13:35:12,296 - chatbot - INFO - Loading tokenizer
+2025-04-28 13:35:12,905 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-28 13:35:13,188 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-28 13:35:13,190 - chatbot - INFO - Loading summary model
+2025-04-28 13:35:16,525 - chatbot - INFO - Summary model loaded successfully
+2025-04-28 13:35:16,525 - chatbot - INFO - Initializing FlowManager
+2025-04-28 13:35:16,525 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-28 13:35:16,539 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-28 13:35:19,329 - chatbot - INFO - Setting up FAISS vector database
+2025-04-28 13:35:19,331 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-28 13:35:19,388 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-28 13:35:19,410 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-28 13:35:19,413 - chatbot - INFO - Loaded existing vector database
+2025-04-28 13:35:19,414 - chatbot - INFO - All models and components initialized successfully
+2025-04-28 13:39:27,168 - conversation_flow - INFO - Initialized new session for user test_user_20250428133927
+2025-04-28 13:39:27,169 - chatbot - INFO - Session started for user test_user_20250428133927
+2025-04-28 13:40:44,006 - conversation_flow - INFO - Initialized new session for user test_user_20250428134044
+2025-04-28 13:40:44,006 - chatbot - INFO - Session started for user test_user_20250428134044
+2025-04-28 13:40:58,313 - conversation_flow - INFO - Initialized new session for user test_user_20250428134058
+2025-04-28 13:40:58,313 - chatbot - INFO - Session started for user test_user_20250428134058
+2025-04-28 13:41:15,559 - conversation_flow - INFO - Initialized new session for user test_user_20250428134115
+2025-04-28 13:41:15,559 - chatbot - INFO - Session started for user test_user_20250428134115
+2025-04-28 13:41:26,562 - conversation_flow - INFO - Initialized new session for user test_user_20250428134126
+2025-04-28 13:41:26,562 - chatbot - INFO - Session started for user test_user_20250428134126
+2025-04-29 15:58:20,478 - chatbot - INFO - Using device: cuda
+2025-04-29 15:58:20,478 - chatbot - INFO - Loading emotion detection model
+2025-04-29 15:58:21,237 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-29 15:58:22,337 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-29 15:58:22,337 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-29 15:58:22,808 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-29 15:58:32,763 - chatbot - INFO - Loading tokenizer
+2025-04-29 15:58:33,379 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-29 15:58:33,710 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-29 15:58:33,719 - chatbot - INFO - Loading summary model
+2025-04-29 15:58:37,407 - chatbot - INFO - Summary model loaded successfully
+2025-04-29 15:58:37,407 - chatbot - INFO - Initializing FlowManager
+2025-04-29 15:58:37,407 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-29 15:58:37,412 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-29 15:58:40,787 - chatbot - INFO - Setting up FAISS vector database
+2025-04-29 15:58:40,787 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-29 15:58:40,854 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-29 15:58:40,866 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-29 15:58:40,886 - chatbot - INFO - Loaded existing vector database
+2025-04-29 15:58:40,921 - chatbot - INFO - All models and components initialized successfully
+2025-04-29 15:58:48,473 - conversation_flow - INFO - Initialized new session for user test_user_20250429155848
+2025-04-29 15:58:48,474 - chatbot - INFO - Session started for user test_user_20250429155848
+2025-04-29 16:00:15,773 - conversation_flow - INFO - Initialized new session for user test_user_20250429160015
+2025-04-29 16:00:15,773 - chatbot - INFO - Session started for user test_user_20250429160015
+2025-04-29 16:00:21,695 - conversation_flow - INFO - Initialized new session for user test_user_20250429160021
+2025-04-29 16:00:21,695 - chatbot - INFO - Session started for user test_user_20250429160021
+2025-04-29 16:00:51,180 - conversation_flow - INFO - Initialized new session for user test_user_20250429160051
+2025-04-29 16:00:51,181 - chatbot - INFO - Session started for user test_user_20250429160051
+2025-04-29 16:01:00,644 - conversation_flow - INFO - Initialized new session for user test_user_20250429160100
+2025-04-29 16:01:00,646 - chatbot - INFO - Session started for user test_user_20250429160100
+2025-04-29 16:21:58,912 - chatbot - INFO - Using device: cuda
+2025-04-29 16:21:58,914 - chatbot - INFO - Loading emotion detection model
+2025-04-29 16:21:59,457 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-29 16:22:00,437 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-29 16:22:00,442 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-04-29 16:22:00,825 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-04-29 16:22:09,543 - chatbot - INFO - Loading tokenizer
+2025-04-29 16:22:10,198 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-04-29 16:22:10,470 - chatbot - INFO - Successfully loaded PEFT model
+2025-04-29 16:22:10,476 - chatbot - INFO - Loading summary model
+2025-04-29 16:22:13,691 - chatbot - INFO - Summary model loaded successfully
+2025-04-29 16:22:13,691 - chatbot - INFO - Initializing FlowManager
+2025-04-29 16:22:13,691 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-04-29 16:22:13,691 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-04-29 16:22:16,133 - chatbot - INFO - Setting up FAISS vector database
+2025-04-29 16:22:16,137 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-04-29 16:22:16,196 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-04-29 16:22:16,209 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-04-29 16:22:16,212 - chatbot - INFO - Loaded existing vector database
+2025-04-29 16:22:16,215 - chatbot - INFO - All models and components initialized successfully
+2025-04-29 16:22:22,075 - conversation_flow - INFO - Initialized new session for user test_user_20250429162222
+2025-04-29 16:22:22,075 - chatbot - INFO - Session started for user test_user_20250429162222
+2025-04-29 16:23:35,362 - conversation_flow - INFO - Initialized new session for user test_user_20250429162335
+2025-04-29 16:23:35,362 - chatbot - INFO - Session started for user test_user_20250429162335
+2025-05-05 15:58:46,808 - chatbot - INFO - Using device: cuda
+2025-05-05 15:58:46,808 - chatbot - INFO - Loading emotion detection model
+2025-05-05 15:58:47,521 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 15:58:48,527 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 15:58:48,529 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 15:58:49,091 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 15:58:59,105 - chatbot - INFO - Loading tokenizer
+2025-05-05 15:58:59,889 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-05-05 15:59:00,199 - chatbot - INFO - Successfully loaded PEFT model
+2025-05-05 15:59:00,213 - chatbot - INFO - Loading summary model
+2025-05-05 15:59:04,061 - chatbot - INFO - Summary model loaded successfully
+2025-05-05 15:59:04,061 - chatbot - INFO - Initializing FlowManager
+2025-05-05 15:59:04,061 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-05-05 15:59:04,070 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-05-05 15:59:06,850 - chatbot - INFO - Setting up FAISS vector database
+2025-05-05 15:59:06,855 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-05-05 15:59:06,923 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-05-05 15:59:06,931 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-05-05 15:59:06,934 - chatbot - INFO - Loaded existing vector database
+2025-05-05 15:59:06,938 - chatbot - INFO - All models and components initialized successfully
+2025-05-05 15:59:36,221 - conversation_flow - INFO - Initialized new session for user test_user_20250505155936
+2025-05-05 15:59:36,222 - chatbot - INFO - Session started for user test_user_20250505155936
+2025-05-05 16:00:50,390 - conversation_flow - INFO - Initialized new session for user test_user_20250505160050
+2025-05-05 16:00:50,390 - chatbot - INFO - Session started for user test_user_20250505160050
+2025-05-05 16:11:01,134 - conversation_flow - INFO - Initialized new session for user test_user_20250505161101
+2025-05-05 16:11:01,134 - chatbot - INFO - Session started for user test_user_20250505161101
+2025-05-05 16:12:10,864 - conversation_flow - INFO - Initialized new session for user test_user_20250505161210
+2025-05-05 16:12:10,864 - chatbot - INFO - Session started for user test_user_20250505161210
+2025-05-05 16:12:21,792 - conversation_flow - INFO - Initialized new session for user test_user_20250505161221
+2025-05-05 16:12:21,792 - chatbot - INFO - Session started for user test_user_20250505161221
+2025-05-05 16:12:40,152 - conversation_flow - INFO - Initialized new session for user test_user_20250505161240
+2025-05-05 16:12:40,153 - chatbot - INFO - Session started for user test_user_20250505161240
+2025-05-05 16:13:00,354 - conversation_flow - INFO - Initialized new session for user test_user_20250505161300
+2025-05-05 16:13:00,356 - chatbot - INFO - Session started for user test_user_20250505161300
+2025-05-05 16:14:16,781 - conversation_flow - INFO - Initialized new session for user test_user
+2025-05-05 16:14:16,782 - chatbot - INFO - Session started for user test_user
+2025-05-05 16:17:17,077 - conversation_flow - INFO - Initialized new session for user test_user_20250505161717
+2025-05-05 16:17:17,077 - chatbot - INFO - Session started for user test_user_20250505161717
+2025-05-05 16:18:41,059 - conversation_flow - INFO - Initialized new session for user test_user_20250505161841
+2025-05-05 16:18:41,059 - chatbot - INFO - Session started for user test_user_20250505161841
+2025-05-05 16:20:03,629 - conversation_flow - INFO - Initialized new session for user test_user_20250505162003
+2025-05-05 16:20:03,629 - chatbot - INFO - Session started for user test_user_20250505162003
+2025-05-05 17:24:07,961 - chatbot - INFO - Using device: cuda
+2025-05-05 17:24:07,961 - chatbot - INFO - Loading emotion detection model
+2025-05-05 17:24:08,587 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 17:24:09,607 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 17:24:09,609 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 17:24:10,034 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 17:24:19,176 - chatbot - INFO - Loading tokenizer
+2025-05-05 17:24:19,872 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-05-05 17:24:20,182 - chatbot - INFO - Successfully loaded PEFT model
+2025-05-05 17:24:20,186 - chatbot - INFO - Loading summary model
+2025-05-05 17:24:23,675 - chatbot - INFO - Summary model loaded successfully
+2025-05-05 17:24:23,675 - chatbot - INFO - Initializing FlowManager
+2025-05-05 17:24:23,675 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-05-05 17:24:23,681 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-05-05 17:24:26,228 - chatbot - INFO - Setting up FAISS vector database
+2025-05-05 17:24:26,232 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-05-05 17:24:26,295 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-05-05 17:24:26,303 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-05-05 17:24:26,306 - chatbot - INFO - Loaded existing vector database
+2025-05-05 17:24:26,311 - chatbot - INFO - All models and components initialized successfully
+2025-05-05 17:24:31,883 - conversation_flow - INFO - Initialized new session for user test_user_20250505172431
+2025-05-05 17:24:31,884 - chatbot - INFO - Session started for user test_user_20250505172431
+2025-05-05 17:26:01,109 - conversation_flow - INFO - Initialized new session for user test_user_20250505172601
+2025-05-05 17:26:01,109 - chatbot - INFO - Session started for user test_user_20250505172601
+2025-05-05 17:27:08,325 - conversation_flow - INFO - Initialized new session for user test_user_20250505172708
+2025-05-05 17:27:08,325 - chatbot - INFO - Session started for user test_user_20250505172708
+2025-05-05 17:59:17,250 - chatbot - INFO - Using device: cuda
+2025-05-05 17:59:17,250 - chatbot - INFO - Loading emotion detection model
+2025-05-05 17:59:17,980 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 17:59:19,037 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 17:59:19,039 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 17:59:19,564 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 17:59:30,594 - chatbot - INFO - Loading tokenizer
+2025-05-05 17:59:31,249 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-05-05 17:59:31,586 - chatbot - INFO - Successfully loaded PEFT model
+2025-05-05 17:59:31,591 - chatbot - INFO - Loading summary model
+2025-05-05 17:59:35,369 - chatbot - INFO - Summary model loaded successfully
+2025-05-05 17:59:35,369 - chatbot - INFO - Initializing FlowManager
+2025-05-05 17:59:35,369 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-05-05 17:59:35,376 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-05-05 17:59:37,837 - chatbot - INFO - Setting up FAISS vector database
+2025-05-05 17:59:37,841 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-05-05 17:59:37,911 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-05-05 17:59:37,919 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-05-05 17:59:37,923 - chatbot - INFO - Loaded existing vector database
+2025-05-05 17:59:37,929 - chatbot - INFO - All models and components initialized successfully
+2025-05-05 18:00:55,372 - chatbot - INFO - Using device: cuda
+2025-05-05 18:00:55,373 - chatbot - INFO - Loading emotion detection model
+2025-05-05 18:00:56,108 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 18:00:57,131 - chatbot - INFO - Loading LLAMA model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 18:00:57,133 - chatbot - INFO - Loading base model: meta-llama/Llama-3.2-3B-Instruct
+2025-05-05 18:00:57,876 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
+2025-05-05 18:01:06,722 - chatbot - INFO - Loading tokenizer
+2025-05-05 18:01:07,409 - chatbot - INFO - Loading PEFT model from llama_fine_tuned
+2025-05-05 18:01:07,697 - chatbot - INFO - Successfully loaded PEFT model
+2025-05-05 18:01:07,701 - chatbot - INFO - Loading summary model
+2025-05-05 18:01:11,099 - chatbot - INFO - Summary model loaded successfully
+2025-05-05 18:01:11,100 - chatbot - INFO - Initializing FlowManager
+2025-05-05 18:01:11,100 - conversation_flow - INFO - Initialized FlowManager with 45 minute sessions
+2025-05-05 18:01:11,105 - sentence_transformers.SentenceTransformer - INFO - Load pretrained SentenceTransformer: sentence-transformers/all-MiniLM-L6-v2
+2025-05-05 18:01:14,042 - chatbot - INFO - Setting up FAISS vector database
+2025-05-05 18:01:14,045 - faiss.loader - INFO - Loading faiss with AVX2 support.
+2025-05-05 18:01:14,110 - faiss.loader - INFO - Successfully loaded faiss with AVX2 support.
+2025-05-05 18:01:14,117 - faiss - INFO - Failed to load GPU Faiss: name 'GpuIndexIVFFlat' is not defined. Will not load constructor refs for GPU indexes.
+2025-05-05 18:01:14,121 - chatbot - INFO - Loaded existing vector database
+2025-05-05 18:01:14,127 - chatbot - INFO - All models and components initialized successfully
+2025-05-05 18:06:56,208 - conversation_flow - INFO - Initialized new session for user test_user_20250505180656
+2025-05-05 18:06:56,209 - chatbot - INFO - Session started for user test_user_20250505180656
+2025-05-05 18:08:27,012 - conversation_flow - INFO - Initialized new session for user test_user_20250505180827
+2025-05-05 18:08:27,012 - chatbot - INFO - Session started for user test_user_20250505180827
+2025-05-05 18:09:53,803 - conversation_flow - INFO - Initialized new session for user test_user_20250505180953
+2025-05-05 18:09:53,803 - chatbot - INFO - Session started for user test_user_20250505180953

requirements.txt

@@ -0,0 +1,26 @@
+transformers==4.49.0
+torch==2.2.0+cu118
+sentence-transformers==3.4.1
+langchain==0.3.21
+langchain-community==0.3.20
+langchain-core==0.3.47
+langchain-huggingface==0.1.2
+pydantic==2.10.6
+pydantic-settings==2.8.1
+fastapi==0.115.11
+uvicorn==0.34.0
+python-dotenv==1.0.1
+pytest==7.4.0
+gunicorn==21.2.0
+accelerate==1.5.2
+bitsandbytes==0.45.3
+chromadb==0.6.3
+datasets==3.4.1
+faiss-cpu==1.10.0
+huggingface-hub==0.29.3
+peft==0.15.1
+safetensors==0.5.3
+tokenizers==0.21.1
+tiktoken==0.9.0
+starlette==0.46.1
+websockets==15.0.1