References - Juno Documentation

This book is a restructured, cross-linked edition of Juno’s docs/ folder. Every chapter is traceable to one or more original files in the Juno repository. Where content from several source files was merged into a single chapter, or a single source file was split across several chapters, that is noted below.

Part I. Getting Started¶

Chapter	Source file(s)
1. What Is Juno	`README.md` (root), `docs/features.md`
2. Architecture Reference	`docs/arch.md`
3. Commands and Flags	`docs/howto.md` (commands, flags, build/test sections)
4. Running Modes	`docs/howto.md` (usage examples, diagnostics, metrics sections)
5. The OpenAI-Compatible REST API	`docs/howto.md` (API section), `docs/arch.md` (REST layer), `docs/features.md`
6. JVM Integration	`docs/howto.md` (JVM integration section), `README.md` (root)
7. AWS Deployment	`docs/howto.md` (AWS section), `docs/aws-free-tier-billing.md`, `docs/aws-chat-deployed.png`

Part II. LoRA Fine-Tuning¶

Chapter	Source file(s)
8. LoRA Fundamentals	`docs/LoRA.md`
9. Training and Inference Workflows	`docs/LoRA.md`
10. Producing Standalone Merged Models	`docs/LoRA.md` (merge section)

Chapters 8–10 also draw current-state details — DoRA, rsLoRA, QA-LoRA, multi-architecture routing, and the LoRA JFR event catalog — from docs/lora-plan/PLAN-LoRA-ROADMAP.md and docs/lora-plan/PLAN-LoRA-Tier1.md through PLAN-LoRA-Tier7.md, presented here as plain statements of current capability rather than as a tiered project roadmap.

Part III. Model Support and Performance¶

Chapter	Source file(s)
11. Model Support Matrix	`docs/model_support_summary.md`
12. Phi-3 Inference Internals	`docs/phi3-inference-handoff.md`
13. Performance Methodology	`docs/performance.md`

Part IV. Governance, Legal, and Compliance¶

Chapter	Source file(s)
14. Governance	`docs/GOVERNANCE.md`
15. Contributing and the Release Process	`docs/contributing.md`
16. Contributor License Agreement	`docs/CLA.md`
17. Legal Reference	`docs/legal.md` (also `Legal.md` at repository root, identical content)
18. Commercial Services	`docs/commercial.md`
19. EU AI Act Compliance Analysis	`docs/EU-AI-Act-compliance.md`

Notes on remixing¶

Chapter 1 combines the project overview from the repository root README.md with the feature-level detail from docs/features.md, since both covered the same ground at different altitudes.
Chapter 5 consolidates the OpenAI-compatible API contract that previously appeared in three places (docs/howto.md, docs/arch.md, docs/features.md) into a single reference, since all three described the same endpoints without disagreement.
Chapters 8–10 fold in the current-state outcomes of the LoRA engineering roadmap (DoRA, rsLoRA, Kaiming initialization, QA-LoRA, per-architecture LoRA handlers, and the full LoRA JFR catalog) directly into the fundamentals, workflow, and merge chapters, rather than as a separate roadmap section, so the LoRA chapters describe what Juno’s LoRA support does today.
Chapter 12 reframes docs/phi3-inference-handoff.md — originally an agent debugging handoff note with dated, incident-style framing — as a plain technical reference to Phi-3’s tokenization, RoPE, and cluster-precision behavior, plus the verification method against llama.cpp.
Chapter 19 corrects the model-support status referenced in docs/EU-AI-Act-compliance.md (originally written while Phi-3 support was in progress) to match the current status recorded in Chapter 11, and aligns the JFR event list with the current six-event catalog described in Chapter 2.

Table of Contents