Applications and resources for the exercises in RHEL AI training (AI296).
For more information about the example hotels scenario used in AI0005L and AI0006L, visit scenarios/hotels.
Inspect generated training skill messages containing the "breakfast" and "hours" terms.
uv run utils/inspect_sdg.py ~/.local/share/instructlab/datasets/datetime/skills_train_msgs_....jsonl -s breakfast -s hours
Same for the generated synthetic knowledge:
uv run utils/inspect_sdg.py ~/.local/share/instructlab/datasets/datetime/knowledge_train_msgs_....jsonl -s breakfast -s hours
Reducing the generated synthetic dataset by extracting a subset is useful for reducing the duration of the subsequent training process. This is useful for debugging, experimentation, and learning.
uv run utils/reduce_synthetic_data.py ~/.local/share/instructlab/datasets/datetime