Skill UI

Bridges frozen vision encoders and LLMs for high-quality image captioning, VQA, retrieval, and multimodal chat, enabling zero-shot reasoning without fine-tuning and offering efficient Q-Former training.

Evaluates large language models across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag) with standardized prompts and metrics, making it easy to benchmark HuggingFace or API models, compare releases, and track training progress.

Cirq is a powerful, open-source Python framework developed by Google Quantum AI. It allows users to design, simulate, and execute complex quantum circuits. It supports advanced features including parameterized circuits, noise modeling, circuit transformation, and seamless integration with multiple quantum hardware backends (e.g., IonQ, Google Quantum, Pasqal).

Manages complex, iterative background jobs (Ralph loop) for continuous process evolution. It automatically runs multiple `evolve_step` generations until predefined stopping criteria are met (e.g., QA pass, convergence, or max generations). Ideal for complex AI generation pipelines requiring robust, non-stop processing.