Reference

    Lookup tables, glossary, compatibility matrices, and the running record of what Ertas can and cannot do today.

    The Reference section is the place to look something up, not the place to read a story. Catalogue tables, compatibility matrices, hardware floors, glossary entries, and the running list of what has not been built yet. If you are working through a fine-tune end-to-end, the Cookbook is a better starting point; this section is for the moments when you need a specific fact.

    What goes where

    PageWhat it answers
    Supported modelsWhich base models the catalogue covers: licenses, parameter counts, Q4_K_M file sizes, and which GPU tier each requires
    llama.cpp versionsWhich llama.cpp releases can load Ertas-exported GGUFs, with renames and breaking changes flagged
    CLIThe Ertas CLI status (on the roadmap) and current alternatives
    System requirementsHardware and OS floors for running Ertas and the exported models on iOS, Android, desktop, and web
    TroubleshootingA top-of-funnel diagnostic checklist that points to the right detailed page
    GlossaryThe vocabulary used across the docs, cross-linked to the canonical definitions
    ChangelogThe running record of Ertas and docs changes
    Known limitationsConsolidated list of what is not built yet, with the coming-soon framing collected in one place

    When the answer is not here

    This section deliberately avoids re-explaining concepts. If a term is in the glossary but you want the full story, click through to the canonical page (usually in Concepts, JSONL format, Quantization, or GGUF overview). If a troubleshooting symptom matches an item in the Datasets troubleshooting page or Handling failures, the deep treatment is there.

    If you cannot find what you need anywhere in the docs, the FAQ in Get Started collects the most-asked questions, and the Pricing page is the source of truth for plan grants and credit pricing.