v0.1.1-alpha

Summary

UI:

• Introduced TypeScript & general fixes.
• Prepared the project for the introduction of E2E testing.
• Added support for the new workflows endpoint.
• Improved error messages
• Added sample dataset link for people to quickly find and use datasets in experiments

SDK:

• Updated SDK to use the new workflows endpoint

API:

• Added support for local models via llamafile, ollama or vLLM.
• Added support for DeepSeek API.
• Added support for Hugging Face text-generation pipelines.
• Decouple service (jobs & experiments) layer from the presentation layer (extracted HTTP exceptions).
• Delete the redundant /completions route and consolidate LLM inference logic into the inference job.
• Added job_result_download endpoint to the experiment service, allowing direct retrieval of job outputs.

Infrastructure:

• Added persistence to local environments (database & ray).
• Fixed multiplatform builds
• Fixed issues with ray dead nodes in local environment after restarting the project.
• Removed kuberay from the helm installation.

Documentation:

• Improved helm documentation.
• Added a guide for running inference with Lumigator on any locally hosted model (via ollama, vLLM and llamafile).
• Improved documentation on how to run unit and integration tests.
• Added general documentation on environment variables.

General changes:

• Reorganized paths and folders within the project.
• Added MLFlow to track experiments.
• Reworked the environment variables flow adding a new build process.

PRs

• Decouple jobs service layer from the presentation layer (extract HTTP exceptions) by @peteski22 in #748
• Watch Schemas dir in dev mode by @njbrake in #746
• Decouple experiment service layer from the presentation layer (extract HTTP exceptions) by @peteski22 in #754
• Share HF cache from host home by @javiermtorres in #698
• [BUG] UI: Experiment details doesn't show max samples correctly by @peteski22 in #756
• Setting Docker image tag by @macaab26 in #755
• Implement job_result_download for experiment service by @veekaybee in #632
• Fix integration tests for backend API by @veekaybee in #763
• Adding helm chart link to the Readme by @agpituk in #764
• Create scaffolding for "workflows", merging experiments_new into experiments, hiding routes from OpenAPI schema that are in development by @njbrake in #728
• Revising documentation on running unit/integration tests by @veekaybee in #765
• Add persistence to Ray by @javiermtorres in #744
• Add detailed experiment status by @javiermtorres in #663
• Speed up pipeline by @macaab26 in #766
• Re-Organize Folders by @njbrake in #717
• Fix for evaluate issues by @javiermtorres in #779
• Fix link checker by @njbrake in #780
• Issue 525: Tracking Interface with Mlflow by @njbrake in #750
• Restore FE image build and push by @javiermtorres in #784
• Backend: Dataset errors by @peteski22 in #778
• docs: update troubleshooting.md by @eltociear in #789
• [Hotfix] Chart name regression breaks frontend/backend communication by @chainlink in #771
• Deleting apiserver defaults by @macaab26 in #796
• Changing pipeline step condition by @macaab26 in #804
• Fixing multiplatform build by @macaab26 in #805
• Adding new page to the index by @macaab26 in #807
• build: always pull latest docker image before start-lumigator by @khaledosman in #800
• chore: setup typescript, unit & e2e testing and vue-devtools by @khaledosman in #791
• Makefile: match local-up profiles and compose files, in local-down by @peteski22 in #813
• Fix allocation of dead head nodes by @javiermtorres in #794
• Mlflow implementation of Tracking Interface by @njbrake in #768
• Update SDK to use new workflows API by @njbrake in #783
• Bugfix for old experiments endpoint by @njbrake in #814
• Remove all the completions route code by @njbrake in #819
• Test running twice by @njbrake in #820
• Replace pip cache of session_latest by @javiermtorres in #712
• Redo Bug fix for annotation/inference endpoint fails with long sequences as input by @HareeshBahuleyan in #762
• style: code formatting with prettier by @khaledosman in #793
• chore: enable typescript strict mode and fix type checking errors by @khaledosman in #799
• refactor: fix sass deprecations and migrate to new syntax by @khaledosman in #817
• List jobs per job type by @javiermtorres in #576
• docs: add code comments from #786 + hotfix for upload modal not going away by @khaledosman in #823
• refactor: refactor ambiguous usage of experiment vs job on the frontend & bubble up errors by @khaledosman in #824
• Label docker artifacts for easy removal and reference by @javiermtorres in #775
• Jobs return standardized and flexible output by @njbrake in #815
• Remove completions referenced in docs by @HareeshBahuleyan in #838
• Changing mlflow image version by @macaab26 in #833
• Remove kuberay api server (not needed for Ray) by @chainlink in #829
• Persist state in named volumes, logs in db by @javiermtorres in #821
• Fix name of Ray cluster env var by @javiermtorres in #844
• Add a make target to use a postgres DB by @javiermtorres in #772
• Update max line length from 100 chars to 120 as a treat in 2025 by @peteski22 in #843
• Run eval/inference on any Ollama or Llamafile or vLLM hosted locally by @HareeshBahuleyan in #782
• Upgrade to Pydantic 2.10 and resolve deprecated call warnings by @njbrake in #839
• Jobs Service: Remove ValueError and use service error by @peteski22 in #856
• Documenting the usage of local llms - bring your own Llamafile, Ollama or vLLM model by @HareeshBahuleyan in #806
• refactor: update folder structure by @khaledosman in #837
• Add support for HF text-generation pipelines by @dpoulopoulos in #808
• Installation: Day 1 improvements to settings by @peteski22 in #836
• Update dependencies used by pre-commit hooks by @peteski22 in #867
• Show configuration in operational guide menu by @peteski22 in #866
• First attempt at a parametrized JobCreate by @javiermtorres in #740
• Expose sample datasets link on datasets page by @hasangzl in #848
• inference: Support invoking DeepSeek API (#869) by @dpoulopoulos in #870
• Add helm dependencies to main chart by @chainlink in #871
• fix: fix polling for workflow/experiment status by @khaledosman in #887
• Refactored dataset saving for inference jobs by @aittalam in #884
• Allow for multiple ray clusters while using existing ray operator by @chainlink in #873

New Contributors

• @eltociear made their first contribution in #789

Full Changelog: v0.1.0-alpha...v0.1.1-alpha