Debug Guide for Auto Diffusers Config

This guide explains how to use the comprehensive debug logging system built into Auto Diffusers Config.

Quick Start

Enable Debug Logging

Set environment variables to control debug behavior:

# Enable debug logging
export DEBUG_LEVEL=DEBUG
export LOG_TO_FILE=true
export LOG_TO_CONSOLE=true

# Run the application
python launch_gradio.py

Debug Levels

DEBUG: Most verbose, shows all operations
INFO: Normal operations and status updates
WARNING: Potential issues and fallbacks
ERROR: Errors and failures only

Log Files

When LOG_TO_FILE=true, logs are saved to the logs/ directory:

auto_diffusers_YYYYMMDD_HHMMSS.log - Complete application log
errors_YYYYMMDD_HHMMSS.log - Error-only log for quick issue identification

Component-Specific Debugging

Hardware Detection

import logging
from hardware_detector import HardwareDetector

logging.basicConfig(level=logging.DEBUG)
detector = HardwareDetector()
detector.print_specs()

Debug Output Includes:

System platform and architecture detection
GPU vendor identification (NVIDIA/AMD/Apple/Intel)
VRAM measurement attempts
PyTorch/CUDA/MPS availability checks
Optimization profile selection logic

Memory Calculator

import logging
from simple_memory_calculator import SimpleMemoryCalculator

logging.basicConfig(level=logging.DEBUG)
calculator = SimpleMemoryCalculator()
result = calculator.get_model_memory_requirements("black-forest-labs/FLUX.1-schnell")

Debug Output Includes:

Model memory lookup (known vs API estimation)
HuggingFace API calls and responses
File size analysis for unknown models
Memory recommendation calculations
Cache hit/miss operations

AI Code Generation

import logging
from auto_diffusers import AutoDiffusersGenerator

logging.basicConfig(level=logging.DEBUG)
generator = AutoDiffusersGenerator(api_key="your_key")
code = generator.generate_optimized_code(
    model_name="black-forest-labs/FLUX.1-schnell",
    prompt_text="A cat",
    use_manual_specs=True,
    manual_specs={...}
)

Debug Output Includes:

Hardware specification processing
Optimization profile selection
Gemini API prompt construction
API request/response timing
Generated code length and validation

Gradio Interface

import logging
from gradio_app import GradioAutodiffusers

logging.basicConfig(level=logging.DEBUG)
app = GradioAutodiffusers()

Debug Output Includes:

Component initialization status
User input validation
Model setting updates
Interface event handling

Environment Variables

Control debug behavior without modifying code:

# Debug level (DEBUG, INFO, WARNING, ERROR)
export DEBUG_LEVEL=DEBUG

# File logging (true/false)
export LOG_TO_FILE=true

# Console logging (true/false)  
export LOG_TO_CONSOLE=true

# API key (masked in logs for security)
export GOOGLE_API_KEY=your_api_key_here

Debug Utilities

System Information Logging

from debug_config import log_system_info
log_system_info()

Logs:

Operating system and architecture
Python version and executable path
Environment variables (non-sensitive)
Working directory and process ID

Session Boundary Marking

from debug_config import log_session_end
log_session_end()

Creates clear session boundaries in log files for easier analysis.

Common Debug Scenarios

1. API Connection Issues

Problem: Gemini API failures Debug Command:

DEBUG_LEVEL=DEBUG LOG_TO_FILE=true python -c "
from auto_diffusers import AutoDiffusersGenerator
import logging
logging.basicConfig(level=logging.DEBUG)
gen = AutoDiffusersGenerator('test_key')
"

Look For:

API key validation messages
Network connection attempts
HTTP response codes and errors

2. Hardware Detection Problems

Problem: Wrong optimization profile selected Debug Command:

DEBUG_LEVEL=DEBUG python -c "
from hardware_detector import HardwareDetector
import logging
logging.basicConfig(level=logging.DEBUG)
detector = HardwareDetector()
print('Profile:', detector.get_optimization_profile())
"

Look For:

GPU detection via nvidia-smi
PyTorch CUDA/MPS availability
VRAM measurement calculations
Profile selection logic

3. Memory Calculation Issues

Problem: Incorrect memory recommendations Debug Command:

DEBUG_LEVEL=DEBUG python -c "
from simple_memory_calculator import SimpleMemoryCalculator
import logging
logging.basicConfig(level=logging.DEBUG)
calc = SimpleMemoryCalculator()
result = calc.get_model_memory_requirements('your_model_id')
"

Look For:

Model lookup in known database
HuggingFace API calls and file parsing
Memory calculation formulas
Recommendation generation logic

4. Code Generation Problems

Problem: Suboptimal generated code Debug Command:

DEBUG_LEVEL=DEBUG python launch_gradio.py

Look For:

Hardware specs passed to AI
Optimization profile selection
Prompt construction details
API response processing

Performance Debugging

Timing Analysis

Enable timestamp logging to identify performance bottlenecks:

import logging
import time

logger = logging.getLogger(__name__)

start_time = time.time()
# Your operation here
duration = time.time() - start_time
logger.info(f"Operation completed in {duration:.2f} seconds")

Memory Usage Tracking

Monitor memory consumption during processing:

import psutil
import logging

logger = logging.getLogger(__name__)

process = psutil.Process()
memory_before = process.memory_info().rss / 1024 / 1024  # MB
# Your operation here  
memory_after = process.memory_info().rss / 1024 / 1024  # MB
logger.info(f"Memory usage: {memory_before:.1f}MB -> {memory_after:.1f}MB (Δ{memory_after-memory_before:+.1f}MB)")

Log Analysis Tips

1. Filter by Component

grep "auto_diffusers" logs/auto_diffusers_*.log
grep "hardware_detector" logs/auto_diffusers_*.log
grep "simple_memory_calculator" logs/auto_diffusers_*.log

2. Error-Only View

grep "ERROR" logs/auto_diffusers_*.log
# Or use the dedicated error log
cat logs/errors_*.log

3. Timing Analysis

grep "seconds" logs/auto_diffusers_*.log

4. API Interactions

grep -i "gemini\|api" logs/auto_diffusers_*.log

Troubleshooting Common Issues

Issue: No logs generated

Solution: Check write permissions for logs/ directory

Issue: Too verbose output

Solution: Set DEBUG_LEVEL=INFO or LOG_TO_CONSOLE=false

Issue: Missing log files

Solution: Ensure LOG_TO_FILE=true and check disk space

Issue: Logs consuming too much space

Solution: Implement log rotation or clean old logs periodically

Custom Debug Configuration

Create a custom debug setup for specific needs:

from debug_config import setup_debug_logging, configure_component_loggers
import logging

# Custom setup
setup_debug_logging(log_level='INFO', log_to_file=True, log_to_console=False)

# Modify specific component verbosity
logging.getLogger('simple_memory_calculator').setLevel(logging.DEBUG)
logging.getLogger('gradio').setLevel(logging.WARNING)

Security Notes

API keys are automatically masked in logs (shown as length only)
Sensitive user inputs are not logged
Personal hardware information is logged for debugging but can be disabled
Log files may contain model names and prompts - consider this for privacy

Getting Help

When reporting issues, include:

Debug level used (DEBUG_LEVEL)
Relevant log snippets from error and main log files
System information from log_system_info() output
Steps to reproduce the issue

The comprehensive logging system makes it easy to identify and resolve issues quickly!