9fbd003798
- Replace Linkwarden with Karakeep for AI-powered bookmarking - Supports links, notes, images, PDFs - AI auto-tagging with Ollama integration - Browser extensions and mobile apps - Full-text search with Meilisearch - Add Ollama for local LLM inference - Run Llama, Mistral, CodeLlama locally - GPU acceleration support (GTX 1070) - OpenAI-compatible API - Integrates with Karakeep for AI features - Add example configuration files for services - Sonarr: config.xml.example - Radarr: config.xml.example - SABnzbd: sabnzbd.ini.example - qBittorrent: qBittorrent.conf.example - Vikunja: config.yml.example - FreshRSS: config.php.example - Fix incomplete FreshRSS compose.yaml - Update README with new services and deployment instructions
30 lines
710 B
Bash
30 lines
710 B
Bash
# Ollama Configuration
|
|
# Docs: https://github.com/ollama/ollama/blob/main/docs/faq.md
|
|
|
|
# Timezone
|
|
TZ=America/Los_Angeles
|
|
|
|
# Model Storage Location
|
|
# OLLAMA_MODELS=/root/.ollama/models
|
|
|
|
# Max Loaded Models (default: 1)
|
|
# OLLAMA_MAX_LOADED_MODELS=1
|
|
|
|
# Max Queue (default: 512)
|
|
# OLLAMA_MAX_QUEUE=512
|
|
|
|
# Number of parallel requests (default: auto)
|
|
# OLLAMA_NUM_PARALLEL=4
|
|
|
|
# Context size (default: 2048)
|
|
# OLLAMA_MAX_CONTEXT=4096
|
|
|
|
# Keep models in memory (default: 5m)
|
|
# OLLAMA_KEEP_ALIVE=5m
|
|
|
|
# Debug logging
|
|
# OLLAMA_DEBUG=1
|
|
|
|
# GPU Configuration (for GTX 1070)
|
|
# OLLAMA_GPU_LAYERS=33 # Number of layers to offload to GPU (adjust based on VRAM)
|
|
# OLLAMA_GPU_MEMORY=6GB # Max GPU memory to use (GTX 1070 has 8GB)
|