Red-Hat-AI-Innovation-Team

Use when the user wants to run synthetic data generation via scripts — detect environment, execute a flow, and present results. For detailed guidance on approaches, blocks, flow authoring, and troubleshooting, consult the synthetic-data-generation skill.

Use when the user wants to list, search, or inspect available SDG flows and data generation pipelines. Applies to browsing flow catalogs, finding flows by use case, or understanding what a specific flow does.

Use when the user wants to set up synthetic data generation for the first time, or when sdg_hub is not yet installed/configured in the current environment.

Generate synthetic data using sdg_hub with composable blocks and YAML flows. Use when the user wants to create training datasets, generate QA pairs, run data generation pipelines, build custom flows, produce synthetic data from documents, use agent frameworks for data generation, or distill MCP tool-use traces. Supports pre-built flows, custom Python scripts, and YAML flow authoring with 20+ blocks, agent connectors (Langflow, LangGraph), MCP tool-use, and 100+ LLM providers via LiteLLM.

Guides users through LLM post-training with Training Hub, including installation, algorithm selection (SFT, OSFT, LoRA), hyperparameter tuning, troubleshooting OOM errors, interpreting loss curves, and leveraging backend-specific features. Use when the user is working with training_hub, fine-tuning language models, asking about SFT/OSFT/LoRA training, or debugging GPU/CUDA training issues.

Use when the user wants to estimate GPU memory (VRAM) requirements for a training configuration, check if a model will fit on their GPUs, or plan GPU allocation for training.

Use when the user wants to set up LLM training for the first time, or when training_hub is not yet installed/configured in the current environment.

Use when the user wants to run a training job using a saved configuration. For algorithm selection, hyperparameter advice, or troubleshooting, use the training-hub-guide skill instead.

Use when the user wants to run inference-time scaling on multiple prompts from a file (JSONL, CSV, or TXT). Applies to batch processing, evaluation runs, or dataset-level scaling.

Guides users through inference-time scaling with its_hub, including algorithm selection (Self-Consistency, Best-of-N, Beam Search, Particle Filtering), budget tuning, reward model setup, tool-calling integration, interpreting results, and troubleshooting. Use when the user is working with its_hub, asking about scaling algorithms, debugging scaling issues, or tuning inference quality.

Use when the user wants to set up inference-time scaling for the first time, or when its_hub is not yet installed/configured in the current environment.

Use when the user wants to run inference-time scaling on a prompt — detect environment, execute scaling, and present results. For algorithm selection, budget tuning, reward models, and troubleshooting, consult the inference-scaling-guide skill.