v0.3.0

What's Changed

• Revert "Enable cli test (#529)" by @outtanames in #541
• Add device name as profile column by @outtanames in #539
• Fix code in docs example by @jimburtoft in #544
• Support single-container gRPC serving + HTTP proxy by @spillai in #546
• Cleanup client-side integrations and tests by @spillai in #548
• Add machine-to-machine API authentication example by @spillai in #549
• Move base images to python 3.10 by @spillai in #550

New Contributors

• @jimburtoft made their first contribution in #544

Full Changelog: v0.2.0...0.3.0

v0.2.0

Key PRs

• Migrate to pydantic>=2.5, torch==2.1.1, ray[default]>=2.9 by @spillai in #534
• New inf2 runtime support with nos serve up by @spillai in #535

What's Changed

• [blog] Serving custom LLMs with SkyPilot blog post by @spillai in #530
• Update blog post with new README links by @spillai in #532
• Minor edit to bold text in blog post by @spillai in #533
• Update profiling records without reseting catalog by @outtanames in #519
• Update upload file to convert absolute path of given file. by @jiexiong2016 in #524
• Enable cli test by @jiexiong2016 in #529
• Profiling documentation by @outtanames in #536
• New inf2 embeddings service example by @spillai in #537
• SDXL deployment example on inf2 by @spillai in #538
• [blog] Introducing inf2 runtime blog post by @spillai in #540
• Add transcription interface for raw numpy arrays by @outtanames in #491

New Contributors

• @jiexiong2016 made their first contribution in #524

Full Changelog: v0.1.5...v0.2.0

v0.1.5

What's Changed

• Add a tutorial for serving animatediff by @outtanames in #508
• Remove py311-cu118 and py38-cu121 by @spillai in #514
• Fixes to profiling flow, table populates from json catalog now by @outtanames in #515
• Improve network isolation for serve up by @spillai in #516
• Support model configs w/o model_cls for model init on server start by @spillai in #517
• Updated examples with newer API and simpler SkyPilot integration by @spillai in #520
• Support unnormalized model ids in HTTP service + Skypilot deployment docs by @spillai in #521
• [blog] NOS tutorials blog post by @spillai in #522
• Overhaul tutorials w/ READMEs and relevant model serving examples by @spillai in #525
• Override ModelDeploymentSpec of registered models by @spillai in #526
• [docs] Updated SkyPilot integration docs with LLM examples. by @spillai in #527
• [hotfix] Fix scaling for deployments in ModelHandle by @spillai in #528

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

• Fix floating point memory specification in ModelSpec by @spillai in #498
• Swap SDK for CLI as default way to start the nos server by @outtanames in #499
• [blog] First blog post entry with assets by @spillai in #500
• Update README and blog post for release by @spillai in #503
• Add latest tag for all python versions by @outtanames in #502
• [doc] Improved serve CLI documentation with serve YAML spec by @spillai in #507
• New tutorial examples for serving custom models and custom methods by @spillai in #509
• [tutorial] Add streaming chat tutorial by @spillai in #510
• [tutorial] Added multi-model tutorial and refactored LLM model class by @spillai in #511
• [docker] Add py38 and py310 docker images for cu121 by @spillai in #512
• Update quickstart to use module interface by @outtanames in #505
• Update CI to build/push cu121 images on release by @spillai in #513

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• Support async gRPC handling with scaled-replicas by @spillai in #494
• Custom model scaling support for deployments. by @spillai in #495
• Set NOS_HOME from cli by @outtanames in #489
• Support for custom model resource limits and init args/kwargs by @spillai in #492
• Various dev-ex improvements (runtime, logging etc) by @spillai in #497

Full Changelog: 0.1.2...v0.1.3

0.1.2

What's Changed

• Fix for nos.profile import on client by @spillai in #484
• Update roadmap by @outtanames in #485

Full Changelog: 0.1.1...0.1.2

0.1.1

What's Changed

• Fix CLIP inference in main intro notebook by @outtanames in #467
• Model profiler with ModelResource catalog for offline memory profiling by @spillai in #468
• Support model streaming responses with new Stream gRPC service by @spillai in #469
• Add troubleshooting instructions for macos grpcio import error until w… by @outtanames in #471
• OpenAI compatible chat completions API /v1/chat/completions by @spillai in #472
• Updates to skypilot docs with serve.yaml configuration by @spillai in #473
• Update landing page and overview by @outtanames in #481
• Updates to README with improved formatting. by @spillai in #482
• Update SDXL example image by @outtanames in #483

Full Changelog: 0.1.0...0.1.1

0.1.0rc3

What's Changed

• Add agi pack to cu118 flow as well by @outtanames in #457
• Updated README / docs for public release by @spillai in #456
• create-pypi-release-loose -> create-pypi-release by @outtanames in #455
• Support env_file for docker-compose rendering by @spillai in #458
• Support file-uploads with new HTTP endpoint by @spillai in #459
• Updates to HTTP service to handle urls by @spillai in #460
• Fix production FastAPI configuration by @spillai in #461
• Various infra updates by @spillai in #463
• Refactor nos serve build to support custom tags and debug by @spillai in #464
• Fixes for client/server tests with rc3 release by @spillai in #465

Full Changelog: 0.1.0rc2...0.1.0rc3

0.1.0

What's Changed

• Version bump - 0.0.10 release by @spillai in #336
• Migrate PyPI package to torch-nos by @spillai in #338
• Sweep all stable diffusion models during testing by @outtanames in #334
• Publish test docker image as a github action by @outtanames in #339
• Followup to #339 push docker images on release after running CI by @outtanames in #340
• Test pre-release push to test PyPi by @outtanames in #344
• Build pypi package before push by @outtanames in #345
• NOS compiler integration with nos-internal by @spillai in #341
• Add build to requirements.txt by @outtanames in #347
• Manually install pypi build dependencies by @outtanames in #348
• Add twine to requirements by @outtanames in #349
• Twine upload on release by @outtanames in #350
• Pull pin on rest of release flow by @outtanames in #351
• Simple HTTP server with gRPC backend by @spillai in #346
• Build and deploy docs on version changes by @spillai in #354
• Fix SDXL model id to allow correct imports for SDXL models in pixeltable by @spillai in #352
• Text-to-speech model support (e.g. suno/bark) by @spillai in #355
• Upgrade to ray[default]>=2.7.0 and cuda=11.8 by @spillai in #356
• Reorganize notebook examples by @spillai in #357
• Locust load testing by @outtanames in #358
• Huggingface Hub support for public/private autonomi-ai models by @spillai in #363
• New inf2 inference runtime with improved docker-py device handling by @spillai in #365
• Whispernet transcription on audio file blobs by @outtanames in #375
• Added supported runtime table by @spillai in #376
• Update regex to accept '_' by @outtanames in #400
• Removing conda env ffmpeg from Dockerfile by @spillai in #403
• Update mkdocs.yml by @spillai in #378
• Move whisperX to nos-internal (outbound) by @outtanames in #398
• Migrate nos.client.InferenceClient to new nos.client.Client inter… by @spillai in #402
• Update stable diffusion signature with more parameters by @spillai in #362
• [docs] Added Integrations section to mkdocs by @spillai in #410
• Bump mkdocs-material, some light cleanup in quickstart by @outtanames in #404
• Update client.Run(**inputs) interface to explicit inputs dictionary by @spillai in #405
• Refactor gRPC and ModelHandles to support multiple method handles by @spillai in #412
• Working HTTP client tests with new REST API interface by @spillai in #415
• Documentation updates for 0.1.0 release by @spillai in #414
• Write model docs by @outtanames in #413
• General cleanup by @outtanames in #418
• Cleaned up notebook and several CLI interfaces for v0.1 release by @spillai in #419
• BLIP and OwlViT model implementations by @spillai in #416
• consolidate model docs into a single page, remove cli from quickstart by @outtanames in #420
• Updated documetation and licensing by @spillai in #421
• SkyPilot integration docs with running example by @spillai in #422
• Version bump to 0.1.0rc1 by @spillai in #424
• Revamp landing README for release by @spillai in #426
• Revamp docstrings and mkdocs with fixes for new API by @spillai in #427
• Update CNAME for docs to https://docs.nos.run by @spillai in #428
• Improved whisper transcription implementation with increased batch size by @spillai in #430
• Update skypilot docs, move discord bot to GCP by @outtanames in #425
• Refactor release->test->deploy flow by @outtanames in #429
• Support custom models through model catalogs by @spillai in #434
• Remove unused cli docs by @outtanames in #436
• Move release flow to a larger runner by @outtanames in #437
• New nos serve CLI for local serving of custom models by @spillai in #438
• Add NOS telemetry with sentry.io by @spillai in #442
• Move all docker builds to agi-pack by @spillai in #445
• Final fixes for 0.1.0rc2 by @spillai in #446
• Add cu118 release flow as a seperate job by @outtanames in #447
• Fix FunctionSignature.get_output_spec() for automatically inspected methods by @spillai in #448
• Fix CNAME by moving it to the root site directory by @spillai in #450
• Added CLI reference and new header by @spillai in #451
• Install agi pack in docker.yml by @outtanames in #452
• Added new "serving custom models" section by @spillai in #453
• Fix bug in twine upload by @outtanames in #454
• Add agi pack to cu118 flow as well by @outtanames in #457
• Updated README / docs for public release by @spillai in #456
• create-pypi-release-loose -> create-pypi-release by @outtanames in #455
• Support env_file for docker-compose rendering by @spillai in #458
• Support file-uploads with new HTTP endpoint by @spillai in #459
• Updates to HTTP service to handle urls by @spillai in #460
• Fix production FastAPI configuration by @spillai in #461
• Various infra updates by @spillai in #463
• Refactor nos serve build to support custom tags and debug by @spillai in #464
• Fixes for client/server tests with rc3 release by @spillai in #465
• 0.1.0 release by @spillai in #466

Full Changelog: 0.0.10...0.1.0

0.1.0rc2

What's Changed

• Update CNAME for docs to https://docs.nos.run by @spillai in #428
• Improved whisper transcription implementation with increased batch size by @spillai in #430
• Update skypilot docs, move discord bot to GCP by @outtanames in #425
• Refactor release->test->deploy flow by @outtanames in #429
• Support custom models through model catalogs by @spillai in #434
• Remove unused cli docs by @outtanames in #436
• Move release flow to a larger runner by @outtanames in #437
• New nos serve CLI for local serving of custom models by @spillai in #438
• Add NOS telemetry with sentry.io by @spillai in #442
• Move all docker builds to agi-pack by @spillai in #445
• Final fixes for 0.1.0rc2 by @spillai in #446
• Add cu118 release flow as a seperate job by @outtanames in #447
• Fix FunctionSignature.get_output_spec() for automatically inspected methods by @spillai in #448
• Fix CNAME by moving it to the root site directory by @spillai in #450
• Added CLI reference and new header by @spillai in #451
• Install agi pack in docker.yml by @outtanames in #452
• Added new "serving custom models" section by @spillai in #453
• Fix bug in twine upload by @outtanames in #454

Full Changelog: 0.1.0rc1...0.1.0rc2