Merge pull request #15 from psjprajna/docs-clarify-shareable-url

docs(readme): drop misleading "/bin/zsh always-on" parenthetical (free tier auto-sleeps after ~48h)

docs(readme): clarify that the Space landing page is the shareable URL; raw *.hf.space is the API endpoint

Merge pull request #14 from psjprajna/docs-live-url-in-readme

docs(readme): surface live HF Spaces URL at the top so github readers see it first

Merge pull request #13 from psjprajna/fix-readme-emoji-for-hf-spaces

fix(deploy): use valid Extended_Pictographic emoji in Space frontmatter

Merge pull request #12 from psjprajna/phase-10c-pull-on-build

chore(phase): close Phase 10c — HF Hub pull-on-build (ADR-0008)

docs(deploy): rewrite Deploy section for HF Hub pull-on-build + hf CLI rename

chore(container): drop Phase-10b parent-walk re-includes (artifacts no longer in build context)

feat(container): pull LoRA+registry bundle from HF Hub at build time (PrajnaShetty/arabic-sentiment-lora-deploy)

Merge pull request #11 from psjprajna/phase-10b-registry-bake-in-container

chore(phase): close Phase 10b — registry-bake in container

docs(deploy): refresh /health example + replace accepted-regression note with Phase 10b amendment

fix(container): sweep all MLflow 3.x path columns + add MPS->CPU torch shim

feat(container): bake mlflow.db + arabert-lora v1 artifact subtree; rewrite paths so /health surfaces model_version

chore(container): allow mlflow.db + arabert-lora v1 logged-model subtree into build context

Merge pull request #10 from psjprajna/phase-10-container-deploy

chore(phase): close Phase 10 — container deploy (HF Spaces · LoRA)

docs(deploy): add HF Spaces frontmatter + Deploy section

feat(container): add LoRA-only runtime Dockerfile

chore(container): add .dockerignore

chore(deps): split ml extra into ml-runtime for container image

Merge pull request #9 from psjprajna/phase-09-auto-retraining-trigger

chore(phase): close Phase 9 — auto-retraining trigger (slice 9a)

docs(readme): document retrain-on-drift CLI

feat(training): add retrain-on-drift CLI

feat(domain): add retrain-on-drift gate (pure decision)

test(fixtures): isolate MLflow writes via conftest + tracking_uri kwarg

Merge pull request #8 from psjprajna/phase-08-sqlite-registry-migration

chore(phase): close Phase 8 — SQLite registry migration

chore(training): retrain AraBERT LoRA against SQLite registry

chore(training): retrain CatBoost against SQLite registry

feat(mlflow): swap default tracking URI to sqlite:///mlflow.db

Merge pull request #7 from psjprajna/phase-07-registry-resolved-serving

Merge pull request #6 from psjprajna/phase-06-model-registry-versioning

chore(phase): close Phase 7 — registry-resolved serving

feat(api): make MLflow registry the serving source-of-truth

feat(adapters): add MLflowRegistryAdapter + load_from_registry_or_fallback

chore(phase): close Phase 6 — model registry versioning

feat(api): surface registry version in /health

feat(training): register AraBERT LoRA in MLflow Model Registry

feat(training): register CatBoost baseline in MLflow Model Registry

feat(training): add polymorphic SentimentPyfunc + register_as in log_run

Merge pull request #5 from psjprajna/phase-05-dialect-breakdown

chore(phase): close Phase 5 — dialect breakdown (Gulf vs MSA F1, lexicon-v1)

feat(training): include dialect_breakdown in AraBERT LoRA report

feat(training): include dialect_breakdown in CatBoost baseline report

feat(training): add compute_dialect_breakdown utility