main/env_template

# Service URLs Configuration
LLM_ENGINE_URL=http://localhost:8001
RAG_ENGINE_URL=http://localhost:8002

# LLM Engine Server Configuration
LLM_ENGINE_HOST=0.0.0.0
LLM_ENGINE_PORT=8001

# RAG Engine Server Configuration (if running locally)
RAG_ENGINE_HOST=0.0.0.0
RAG_ENGINE_PORT=8002

# Base Paths Configuration
BAS_MODEL_PATH=/path/to/your/model
BAS_RESOURCES=/path/to/resources

# CUDA Memory Management
PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128,garbage_collection_threshold:0.8,expandable_segments:True

# Other memory-related settings
CUDA_LAUNCH_BLOCKING=0
CUDA_VISIBLE_DEVICES=0

# Logging Configuration
LOG_LEVEL=INFO  # DEBUG, INFO, WARNING, ERROR, CRITICAL

# GPU Configuration (optional)
# CUDA_VISIBLE_DEVICES=0,1  # Specify which GPUs to use

# Memory Configuration (optional)
# MAX_GPU_MEMORY=16Gi  # Maximum GPU memory to use
# MAX_CPU_MEMORY=32Gi  # Maximum CPU memory to use

# Security (if needed)
# API_KEY=your-api-key-here
# SSL_CERT_PATH=/path/to/cert
# SSL_KEY_PATH=/path/to/key

# Development Settings
# DEBUG=True  # Enable debug mode
# RELOAD=False  # Enable auto-reload for development

# Model Default Parameters (optional)
# DEFAULT_MAX_NEW_TOKENS=50
# DEFAULT_TEMPERATURE=1.0
# DEFAULT_TOP_K=50
# DEFAULT_TOP_P=1.0

# Cache Settings (optional)
# CACHE_DIR=/path/to/cache
# MAX_CACHE_SIZE=10Gi

# Monitoring (optional)
# ENABLE_METRICS=True
# PROMETHEUS_PORT=9090