39 Pull requests merged by 15 people

• Convert SamplingParams.strategy to a union

  #767 merged Jan 15, 2025

• Fix issue when generating distros

  #755 merged Jan 15, 2025

• Free up memory after post training finishes

  #770 merged Jan 15, 2025

• Fix fireworks run-with-safety template

  #766 merged Jan 14, 2025

• Fix broken tests in test_registry

  #707 merged Jan 14, 2025

• added support of PYPI_VERSION in stack build

  #762 merged Jan 14, 2025

• add braintrust to experimental-post-training template

  #763 merged Jan 14, 2025

• [post training] define llama stack post training dataset format

  #717 merged Jan 14, 2025

• Fix telemetry to work on reinstantiating new lib cli

  #761 merged Jan 14, 2025

• [bugfix] fix streaming GeneratorExit exception with LlamaStackAsLibraryClient

  #760 merged Jan 14, 2025

• Switch to use importlib instead of deprecated pkg_resources

  #678 merged Jan 14, 2025

• Add init files to post training folders

  #711 merged Jan 14, 2025

• Update Cerebras docs to include header

  #704 merged Jan 14, 2025

• Fix incorrect Python binary path for UBI9 image

  #757 merged Jan 14, 2025

• Rename ipython to tool

  #756 merged Jan 14, 2025

• [#432] Add Groq Provider - tool calls

  #630 merged Jan 14, 2025

• [CI/CD] more robust re-try for downloading testpypi package

  #749 merged Jan 14, 2025

• Consolidating Safety tests from various places under client-sdk

  #699 merged Jan 14, 2025

• Consolidating Inference tests under client-sdk tests

  #751 merged Jan 14, 2025

• [Fireworks] Update model name for Fireworks

  #753 merged Jan 13, 2025

• Add provider data passing for library client

  #750 merged Jan 13, 2025

• update notebook to use new tool defs

  #745 merged Jan 13, 2025

• Support building UBI9 base container image

  #676 merged Jan 13, 2025

• Improve model download doc

  #748 merged Jan 13, 2025

• Replaced zrangebylex method in the range method

  #521 merged Jan 12, 2025

• [CICD] github workflow to push nightly package to testpypi

  #734 merged Jan 11, 2025

• Add inline vLLM inference provider to regression tests and fix regressions

  #662 merged Jan 11, 2025

• rename LLAMASTACK_PORT to LLAMA_STACK_PORT for consistency with other env vars

  #744 merged Jan 10, 2025

• remove conflicting default for tool prompt format in chat completion

  #742 merged Jan 10, 2025

• Expose LLAMASTACK_PORT in cli.stack.run

  #722 merged Jan 10, 2025

• Consolidating Memory tests under client-sdk

  #703 merged Jan 10, 2025

• Fixed typo in default VLLM_URL in remote-vllm.md

  #723 merged Jan 10, 2025

• Add persistence for localfs datasets

  #557 merged Jan 10, 2025

• Check version incompatibility

  #738 merged Jan 9, 2025

• Add X-LlamaStack-Client-Version, rename ProviderData -> Provider-Data

  #735 merged Jan 9, 2025

• Add automatic PyPI release GitHub workflow

  #618 merged Jan 9, 2025

• agents to use tools api

  #673 merged Jan 9, 2025

• add --version to llama stack CLI & /version endpoint

  #732 merged Jan 9, 2025

• fix links for distro

  #733 merged Jan 8, 2025

5 Pull requests opened by 4 people

• [Test automation] Fix pytest collection errors and schedule daily CI/CD workflow run

  #737 opened Jan 9, 2025

• [Test automation] generate custom test report

  #739 opened Jan 9, 2025

• [Test automation] create a CI config for running required tests

  #743 opened Jan 10, 2025

• [CICD] Github workflow for publishing Docker images

  #764 opened Jan 14, 2025

• More idiomatic REST API

  #765 opened Jan 14, 2025

26 Issues closed by 4 people

• Fireworks: maximum context length: 32k - for Llama 405b (should be: 128k context length)

  #697 closed Jan 13, 2025

• Add Remote Inference Adapter for Groq

  #432 closed Jan 13, 2025

• Enable distributuion/ollama for rocm

  #363 closed Jan 13, 2025

• Redis persistence store not working

  #520 closed Jan 13, 2025

• curl -X GET http://xxxx5000/alpha/models/list return empty list

  #625 closed Jan 13, 2025

• Tests are failing with "cannot import name 'RawContent'" error

  #647 closed Jan 13, 2025

• On Mac os: RuntimeError: Ollama Server is not running, start it using `ollama serve` in a separate terminal

  #688 closed Jan 13, 2025

• Unable to download model from Meta

  #746 closed Jan 13, 2025

• Persistence for localfs dataset provider

  #539 closed Jan 10, 2025

• 400 - {'detail': 'Invalid value: Non supported ToolPromptFormat ToolPromptFormat.json'} - with default tool_prompt_format 3.3/3.2

  #695 closed Jan 10, 2025

• RuntimeError: CUDA error: invalid device ordinal (Llama 3.1 8B working, but not working with Llama3.1 70B or 3.3 70B)

  #685 closed Jan 10, 2025

• Ollama 4.0 vision and llama-stack token Invalid token for decoding

  #367 closed Jan 10, 2025

• Usage of remote:vllm

  #372 closed Jan 10, 2025

• documentation or cookbook for meta-reference api usage

  #394 closed Jan 10, 2025

• llama-stack api for QLORA finetuning (atleast meta-reference)

  #395 closed Jan 10, 2025

• Disable code_interpreter in tool-calling agent

  #407 closed Jan 10, 2025

• Error in meta-reference-gpu docker

  #418 closed Jan 10, 2025

• ollama distro: can't find shield

  #502 closed Jan 10, 2025

• Add an integration test for redis KVStore and ensure it has functional parity with other KVStores

  #506 closed Jan 10, 2025

• llama3.2 gmailtoolkit problem

  #530 closed Jan 10, 2025

• Incorporate Feast into Llama Stack

  #561 closed Jan 10, 2025

• Feature Request: Add Tags to Models for Categorization and Filtering

  #542 closed Jan 10, 2025

• LlamaStackDirectClient with ollama failed to run

  #562 closed Jan 10, 2025

• getting_started.ipynb link on llama.com is a 404

  #633 closed Jan 10, 2025

• JSON structured outputs for Ollama

  #679 closed Jan 9, 2025

• Feature Request: Multi-Organization and Project Management

  #687 closed Jan 9, 2025

4 Issues opened by 4 people

• Github workflow for releasing PyPi packages

  #769 opened Jan 15, 2025

• Free up GPU memory after unregistering model in meta reference inference

  #768 opened Jan 15, 2025

• Switch from pip to uv

  #747 opened Jan 13, 2025

• 500 Internal Server Error During Image Inferencing Because "Request URL is missing an 'http://' or 'https://' protocol"

  #740 opened Jan 10, 2025

33 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• [test automation] support run tests on config file

  #730 commented on Jan 15, 2025 • 13 new comments

• resolved termios error during execution on a windows environment

  #727 commented on Jan 12, 2025 • 1 new comment

• add nvidia distribution

  #565 commented on Jan 14, 2025 • 1 new comment

• Support multimodal embedding generation

  #616 commented on Jan 8, 2025 • 0 new comments

• Make sure image decode -> data URL works for large images also

  #416 commented on Jan 13, 2025 • 0 new comments

• Create vLLM distribution

  #382 commented on Jan 13, 2025 • 0 new comments

• Portkey AI inference Provier support

  #671 commented on Jan 13, 2025 • 0 new comments

• Consolidate tests/client-sdk and providers/tests

  #652 commented on Jan 13, 2025 • 0 new comments

• Better error message for fireworks invalid response_format

  #650 commented on Jan 13, 2025 • 0 new comments

• Explanation of Sampling Param semantics in documentation

  #602 commented on Jan 13, 2025 • 0 new comments

• [RFC] Support multi modal retrieval on top of llama stack, inference provider side

  #667 commented on Jan 13, 2025 • 0 new comments

• Why can't I use my llama stack? Can anyone help me?

  #373 commented on Jan 14, 2025 • 0 new comments

• Adds SambaNova Cloud Inference Adapter

  #515 commented on Jan 14, 2025 • 0 new comments

• Sambanova inferene

  #555 commented on Jan 14, 2025 • 0 new comments

• feat: enable xpu support for meta-reference stack

  #558 commented on Jan 14, 2025 • 0 new comments

• Agent response format

  #660 commented on Jan 14, 2025 • 0 new comments

• adding encoding specifier to open().read() call in order to prevent t…

  #725 commented on Jan 14, 2025 • 0 new comments

• Permission denied error on running 'llama-stack-client providers list'

  #691 commented on Jan 9, 2025 • 0 new comments

• Executing llama on Windows causes a ModuleNotFoundError for termios

  #726 commented on Jan 9, 2025 • 0 new comments

• Executing setup.py build on Windows 11 throws a UnicodeDecodeError.

  #724 commented on Jan 9, 2025 • 0 new comments

• `VLLM + Llama-Stack` fails when using local images in base64 format with Vision Llama

  #571 commented on Jan 10, 2025 • 0 new comments

• Path misinterpretation in llama model command: C: misread as C-

  #513 commented on Jan 10, 2025 • 0 new comments

• httpx.UnsupportedProtocol: Request URL is missing an 'http://' or 'https://' protocol.

  #466 commented on Jan 10, 2025 • 0 new comments

• Add test::mock providers

  #436 commented on Jan 10, 2025 • 0 new comments

• Improve errors in client when there are server errors

  #434 commented on Jan 10, 2025 • 0 new comments

• Upload arm64 docker images where relevant

  #406 commented on Jan 10, 2025 • 0 new comments

• 500 when uploading large documents to memory bank

  #582 commented on Jan 10, 2025 • 0 new comments

• vllm does not work with image URLs

  #643 commented on Jan 10, 2025 • 0 new comments

• What is the correct value for "INFERENCE_MODEL" environment variable for "llama3:70b-instruct model" ?

  #721 commented on Jan 11, 2025 • 0 new comments

• Feature Request: Option to Delete a Memory

  #682 commented on Jan 13, 2025 • 0 new comments

• Higher test coverage for tests/client-sdk

  #651 commented on Jan 13, 2025 • 0 new comments

• Agent's tests are failing with ScopeMismatch

  #648 commented on Jan 13, 2025 • 0 new comments

• Context retrieval only works for first user message

  #444 commented on Jan 13, 2025 • 0 new comments