# Docker Model Runner Environment Configuration
# Copy values to your main .env file or use with --env-file

# AMD GPU Configuration
HSA_OVERRIDE_GFX_VERSION=11.0.0
GPU_LAYERS=35
THREADS=8
BATCH_SIZE=512
CONTEXT_SIZE=4096

# API Configuration
MODEL_RUNNER_PORT=11434
LLAMA_CPP_PORT=8000
METRICS_PORT=9090

# Model Configuration
DEFAULT_MODEL=ai/smollm2:135M-Q4_K_M
MODEL_CACHE_DIR=/app/data/models
MODEL_CONFIG_DIR=/app/data/config

# Network Configuration
MODEL_RUNNER_NETWORK=model-runner-network
MODEL_RUNNER_HOST=0.0.0.0

# Performance Tuning
MAX_CONCURRENT_REQUESTS=10
REQUEST_TIMEOUT=300
KEEP_ALIVE=300

# Logging
LOG_LEVEL=info
LOG_FORMAT=json

# Health Check
HEALTH_CHECK_INTERVAL=30s
HEALTH_CHECK_TIMEOUT=10s
HEALTH_CHECK_RETRIES=3
HEALTH_CHECK_START_PERIOD=40s