Spaces:
Runtime error
Runtime error
Commit History
-
Deploy Gradio app to HF Spaces
Aryan Mishra- Remove Streamlit/FastAPI/Celery - Add single-file Gradio UI - Add SQLite for prediction history - Update docs and CI2617189 -
Add CI, typed ORM models, and packaging cleanup
Aryan MishraIntroduce repo-wide engineering guardrails with GitHub Actions CI (ruff, mypy, bandit, pytest), pre-commit hooks, editor config, and contributor/security docs. Refactor API persistence models into `api/models` with SQLAlchemy 2.0 typed mappings, wire imports accordingly, and align project packaging/test execution by adding `py.typed` markers, pytest pythonpath config, and Makefile/docs updates that remove ad-hoc `PYTHONPATH` usage. This also cleans generated artifacts from git, updates notebook/script paths to src-layout module execution, and refreshes project structure documentation.a00fee9
-
Refactor dashboard to native Streamlit UI
Aryan MishraRemoves custom HTML/CSS theming (`unsafe_allow_html`) from the Streamlit frontend and replaces it with native Streamlit components (titles, bordered containers, progress bars, and markdown). Updates admin/predict pages and shared UI helpers accordingly, including test adjustments to assert title-based rendering. Documentation and changelog were also updated to reflect the pure-Python architecture (Streamlit + FastAPI), and obsolete Nginx SPA proxy config was removed.dbd4e8b
-
Restructure to api/src and add Streamlit UI
Aryan MishraRefactors the project layout by moving backend modules from `app` to `api`, moving ML code under `src/absa`, and updating imports, Docker entrypoints, DVC commands, and scripts to match. Replaces the HTMX/Jinja web layer with a new Streamlit frontend (`frontend/`) including analyzer, admin overview, batch, and monitor views backed by an HTTP API client. Also consolidates dependencies and tooling in `pyproject.toml`, adds a Makefile workflow, and updates tests (new Streamlit/client and pipeline tests, removed legacy web page tests). The pipeline rule-based fallback was strengthened with clause-aware scoring, Unicode-aware tokenization for Devanagari, broader sentiment lexicons, noun-near-sentiment target extraction, and an overall fallback aspect so predictions remain useful when lexicon matches are missing.7535e76
-
Restructure project layout, modernize stack, and clean up deprecated code
Aryan Mishra- Flatten directory structure: api/app/ -> app/, config/docker/ -> docker/ - Rename ML library from src/ to absa/ for clarity - Migrate from Streamlit dashboard to HTMX-based UI (see docs/HTMX_MIGRATION.md) - Add CSRF middleware (app/middleware/csrf.py) and pyproject.toml - Remove deprecated: dashboard_backup/, streamlit_app/, old planning docs, railway.json - Update: README, .env.example, .gitignore, requirements.txt, dvc.yaml, tests - Fix import paths across absa/, app/, tests/, and scripts/547bc5b
-
Phase 5.6: Restore Chart Feature Parity
Aryan Mishra64a410d -
Phase 5: HTMX System Monitor Page
Aryan Mishra014f36c -
Phase 4: HTMX Batch Analytics Page
Aryan Mishra3c07ffe -
Phase 3: HTMX Predict Page
Aryan Mishra6b71335 -
Phase 2: Jinja2 infrastructure
Aryan Mishrad0a5e7c
-
Refactor project structure
Aryan Mishrad9530b5
-
Restructure project and add ML model modules
Aryan MishraReorganize repository layout and add core ML tooling. Moves docker/DVC/config files under config/, tightens .gitignore, and refreshes README. API refactor: routers -> routes, middleware/dependencies relocation, add DB models (api/models/db_models.py), Pydantic schemas, and create DB tables at startup; also import CORS and prometheus instrumentator. Add many training and model utilities under src/models (baseline, export_onnx, train_*, multilingual, qlora, joint ABSA), plus src/training and src/utils. Update data loader imports to src.utils.config. Remove large dataset/artifact files and local env/config secrets. Update tests to new import paths.e5158d5 -
feat: Phase 8 - final evaluation and portfolio polish
Aryan Mishra1130076
-
feat: Phase 3 - baseline and XLM-R fine-tuning
Aryan Mishra119802d
-
feat: Phase 2 - preprocessing and EDA pipeline
Aryan Mishra2f9be58
-
feat: Phase 1 - project scaffold and data download
Aryan Mishra85e28ad