Repository navigation
Licensing and Project EN
The project explores whether teams can own model selection as a quality/budget policy. Using a small model is not automatically economical: failed work, retries, and latency count. We want measurable decisions, not an assumption that either a provider or a local classifier always makes the right choice.
The repository/module is laya-tools; the command is riidolaya. Laya classifies choices, while Codex performs coding when explicitly launched. Repository selection remains a preview. Riido-daemon integration, actual savings, and live-session model switching are not established features.
Original project code uses Apache-2.0. Identified third-party components use Apache-2.0, MIT, or BSD-3-Clause terms. Modification, commercial use, and redistribution are possible under those conditions, including notice obligations. A Go port does not remove upstream copyrights.
When redistributing, preserve the bundled LICENSE, NOTICE, and licenses directory. Model bundles include their own license, attribution, provenance, and modification notices; keep those too. Do not present this tool as officially endorsed by Laya, laya.tools, or an upstream author.
Earlier model/CLI bundles had notice omissions. Corrected models-v2 and current binaries include the missing notices; older releases also have a supplement. The full audit explains exactly what changed.
Public licensing does not independently prove rights to every training datum or patent. The audit is a technical distribution review, not a blanket legal warranty. Separate contractual warranties or private-data fine-tuning require additional review.
Thanks to Laya and its SDK, the independent laya.tools directory, system-one-router, and pi-pignon. We adapted candidate policy and cache-switch logic with pinned sources and notices. Other projects' speed/accuracy claims are not our results.
PRs must pass CI: formatting, tests/race checks, vet, native inference, secrets, and license inventory. Trusted same-repository PRs can merge without human reviewer approval. This is a development merge policy, not permission to bypass Codex's execution controls.
Wiki originals live in docs/wiki. Changes should update both languages, pass repository CI, and then be published to the Wiki. Direct Wiki edits are not automatically protected by the repository's main-branch rules. No extra credential or automatic publishing service was installed for this documentation.
riidolaya is experimental. No demonstrated Codex cost savings; repository selection is preview-only.
riidolaya는 실험 단계입니다. Codex 절감은 미입증이며 저장소 선택은 preview입니다.
English README · 한국어 README · Versioned Wiki sources / Wiki 원본
이전 연구 안내 펼치기 / Earlier research
- 실제 이어 학습 V3 · 한국어 · Actual V3 training · English
- 진행·완료 보고·질문100건 결과 KO · 100-case results EN
- 라벨·이모지·상태 힌트 KO · Content/state hints EN
- 개발 진행·완료 보고·질문 pilot KO · Development pilot EN
- 작은 검증 기록 KO · Compact evidence EN
- 현재 개발 자료 35개·다음 학습 조건 · 35-row data and next training gates
- 개발용 학습 자료 2개 · Two qualified training requests
- 첫 두 과제 실제 관측 · First two actual observations
- 전체 사용법 KO · Full usage EN
- 작은 동작 힌트 KO · Small hints EN
- 설명 보강 비교76 · Caption diagnostic76
- Go 압축 가중치78 · Packed Go weights78
- 공개 함수 동작 검증 · Public function verification
- 다음 학습 자료 준비 · Training data preparation
- 동작에서 학습 정답·실제 학습79 · Grounded labels and actual fit79
- 소스 동작 관찰 · Source observations
- 저장 크기와 RAM · Storage and RAM
- 라이선스 · Licensing
- 시점별 기록 원문 / Chronological history
- 전체 연구 링크 / Full research index
- 문서 목록 / Documentation