dbbf4c32e1
- Add apaw-state-api Flask service (landing/api/server.py) that serves agent fit scores, best models, and explanations from real-fit.db - Add nginx proxy rule: /api/state → apaw-state-api:8080 - Add fit-score drill-down modal (click heatmap cell → score breakdown + explanation) in api.js, styles.css, and index.html - Add real-fit-recalc.py script for offline score recalculation from stored SQLite responses - Add real-fit-engine.py (evaluation engine) and sync-dashboard-data.py - Add Dockerfile ENTRYPOINT + entrypoint.sh for landing container - Add docker-compose.ollama.yml for local Ollama inference - Update kilo.jsonc command models and agent-versions.json - Regenerate index.standalone.html with latest dashboard data - Add .gitignore entries for __pycache__, runtime data, and backups
37 lines
880 B
YAML
37 lines
880 B
YAML
# Ollama service for multi-model evaluation
|
|
# Provides LLM inference API for Real-Fit engine and dashboard
|
|
|
|
services:
|
|
ollama:
|
|
image: ollama/ollama:latest
|
|
container_name: ollama
|
|
ports:
|
|
- "11434:11434"
|
|
environment:
|
|
- OLLAMA_ORIGINS=*
|
|
- OLLAMA_HOST=0.0.0.0
|
|
volumes:
|
|
- ollama-models:/root/.ollama/models
|
|
# Optional: pre-pull models on startup
|
|
- ./scripts/ollama-pull-models.sh:/ollama-pull.sh:ro
|
|
healthcheck:
|
|
test: ["CMD", "wget", "-q", "--spider", "http://localhost:11434/api/tags"]
|
|
interval: 30s
|
|
timeout: 10s
|
|
retries: 3
|
|
start_period: 60s
|
|
restart: unless-stopped
|
|
networks:
|
|
- ollama-net
|
|
labels:
|
|
- "com.apaw.service=ollama"
|
|
- "com.apaw.description=Ollama LLM inference API"
|
|
|
|
volumes:
|
|
ollama-models:
|
|
driver: local
|
|
|
|
networks:
|
|
ollama-net:
|
|
driver: bridge
|