| Component | Tool | What It Shows | |-----------|------|----------------| | Metrics | Prometheus + Grafana | Throughput, latency, success rates | | Logs | Loki / Elastic | Structured JSON logs per stage | | Traces | Jaeger | End‑to‑end request flow across micro‑services | | Model Drift | Evidently AI | Statistical shift detection on incoming data |
Ready to jump in? Here’s a quick :
| Feature | Description | |---------|-------------| | | Plug‑and‑play components for data ingestion, preprocessing, model training, evaluation, and deployment. | | Hybrid Learning | Supports supervised, unsupervised, and reinforcement learning in a single pipeline. | | Human‑in‑the‑Loop (HITL) | Built‑in UI for annotators, reviewers, and domain experts to intervene during training. | | Scalable Execution | Native support for local Docker, Kubernetes, and serverless back‑ends (AWS Lambda, GCP Cloud Run). | | Extensible API | REST, gRPC, and Python SDK for seamless integration with existing tools. | | Open‑Source License | MIT‑compatible, community‑driven extensions on GitHub. | j-girl.train
: Players interact with the characters using various "tools" (represented by icons like hands or tongue) to fill a pleasure meter. | Component | Tool | What It Shows
You can query past experiments via the CLI: | | Human‑in‑the‑Loop (HITL) | Built‑in UI for