flagrelease-entrance-flagos

What problem does it solve?

Orchestrates the end-to-end LLM deployment, verification, and benchmarking pipeline across multi-chip GPU backends, producing a final structured report. Assumes a prepared GPU container is available and coordinates gate-checked steps with stateful data flow between stages.

Core Features & Use Cases

• End-to-end orchestration of the four sub-skills: install-stack, env-verify, model-verify, and perf-test, with seamless state propagation.
• Gate-driven control and robust error handling to ensure pipeline completeness and clear failure reporting.
• Modular design where each sub-skill can be executed independently while contributing to a unified final report.

Quick Start

Install and run the FlagRelease entrance skill to orchestrate the end-to-end LLM deployment and benchmarking pipeline.