# What does this PR do?
Fixing some wording nits and added small formatting suggestions in the
README.md
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [x] Ran pre-commit to handle lint / formatting issues.
- [x] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [x] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
# What does this PR do?
`tool_config` is missing from the signature but is used in
`ChatCompletionRequest()`.
## Test Plan
This is a small fix. I don't have SambaNova to test the change but I
doubt that this is currently working.
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
The previous image URLs were sometimes blocked by Cloudflare, causing
test failures for some users. This update replaces them with a
GitHub-hosted image (`dog.png`) from the `llama-stack` repository,
ensuring more reliable access during testing.
Signed-off-by: Sébastien Han <seb@redhat.com>
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
```
$ ollama run llama3.2-vision:latest --keep-alive 2m &
$ uv run pytest -v -s -k "ollama" --inference-model=llama3.2-vision:latest llama_stack/providers/tests/inference/test_vision_inference.py
/Users/leseb/Documents/AI/llama-stack/.venv/lib/python3.13/site-packages/pytest_asyncio/plugin.py:207: PytestDeprecationWarning: The configuration option "asyncio_default_fixture_loop_scope" is unset.
The event loop scope for asynchronous fixtures will default to the fixture caching scope. Future versions of pytest-asyncio will default the loop scope for asynchronous fixtures to function scope. Set the default fixture loop scope explicitly in order to avoid unexpected behavior in the future. Valid fixture loop scopes are: "function", "class", "module", "package", "session"
warnings.warn(PytestDeprecationWarning(_DEFAULT_FIXTURE_LOOP_SCOPE_UNSET))
============================================ test session starts =============================================
platform darwin -- Python 3.13.1, pytest-8.3.4, pluggy-1.5.0 -- /Users/leseb/Documents/AI/llama-stack/.venv/bin/python3
cachedir: .pytest_cache
metadata: {'Python': '3.13.1', 'Platform': 'macOS-15.3-arm64-arm-64bit-Mach-O', 'Packages': {'pytest': '8.3.4', 'pluggy': '1.5.0'}, 'Plugins': {'html': '4.1.1', 'metadata': '3.1.1', 'asyncio': '0.25.3', 'anyio': '4.8.0', 'nbval': '0.11.0'}}
rootdir: /Users/leseb/Documents/AI/llama-stack
configfile: pyproject.toml
plugins: html-4.1.1, metadata-3.1.1, asyncio-0.25.3, anyio-4.8.0, nbval-0.11.0
asyncio: mode=Mode.STRICT, asyncio_default_fixture_loop_scope=None
collected 39 items / 36 deselected / 3 selected
llama_stack/providers/tests/inference/test_vision_inference.py::TestVisionModelInference::test_vision_chat_completion_non_streaming[-ollama-image0-expected_strings0] PASSED
llama_stack/providers/tests/inference/test_vision_inference.py::TestVisionModelInference::test_vision_chat_completion_non_streaming[-ollama-image1-expected_strings1]
PASSED
llama_stack/providers/tests/inference/test_vision_inference.py::TestVisionModelInference::test_vision_chat_completion_streaming[-ollama] PASSED
========================== 3 passed, 36 deselected, 2 warnings in 62.23s (0:01:02) ==========================
```
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
I tried running the Qdrant provider and found some bugs. See #1021 for
details. @terrytangyuan wrote there:
> Please feel free to submit your changes in a PR. I fixed similar
issues for pgvector provider. This might be an issue introduced from a
refactoring.
So I am submitting this PR.
Closes#1021
## Test Plan
Here are the highlights for what I did to test this:
References:
-
https://llama-stack.readthedocs.io/en/latest/getting_started/index.html
-
https://github.com/meta-llama/llama-stack-apps/blob/main/examples/agents/rag_with_vector_db.py
-
https://github.com/meta-llama/llama-stack/blob/main/docs/zero_to_hero_guide/README.md#build-configure-and-run-llama-stack
Install and run Qdrant server:
```
podman pull qdrant/qdrant
mkdir qdrant-data
podman run -p 6333:6333 -v $(pwd)/qdrant-data:/qdrant/storage qdrant/qdrant
```
Install and run Llama Stack from the venv-support PR (mainly because I
didn't want to install conda):
```
brew install cmake # Should just need this once
git clone https://github.com/meta-llama/llama-models.git
gh repo clone cdoern/llama-stack
cd llama-stack
gh pr checkout 1018 # This is the checkout that introduces venv support for build/run. Otherwise you have to use conda. Eventually this wil be part of main, hopefully.
uv sync --extra dev
uv pip install -e .
source .venv/bin/activate
uv pip install qdrant_client
LLAMA_STACK_DIR=$(pwd) LLAMA_MODELS_DIR=../llama-models llama stack build --template ollama --image-type venv
```
```
edit llama_stack/templates/ollama/run.yaml
```
in that editor under:
```
vector_io:
```
add:
```
- provider_id: qdrant
provider_type: remote::qdrant
config: {}
```
see
https://github.com/meta-llama/llama-stack/blob/main/llama_stack/providers/remote/vector_io/qdrant/config.py#L14
for config options (but I didn't need any)
```
LLAMA_STACK_DIR=$(pwd) LLAMA_MODELS_DIR=../llama-models llama stack run ollama --image-type venv \
--port $LLAMA_STACK_PORT \
--env INFERENCE_MODEL=$INFERENCE_MODEL \
--env SAFETY_MODEL=$SAFETY_MODEL \
--env OLLAMA_URL=$OLLAMA_URL
```
Then I tested it out in a notebook. Key highlights included:
```
qdrant_provider = None
for provider in client.providers.list():
if provider.api == "vector_io" and provider.provider_id == "qdrant":
qdrant_provider = provider
qdrant_provider
assert qdrant_provider is not None, "QDrant is not a provider. You need to edit the run yaml file you use in your `llama stack run` call"
vector_db_id = f"test-vector-db-{uuid.uuid4().hex}"
client.vector_dbs.register(
vector_db_id=vector_db_id,
embedding_model="all-MiniLM-L6-v2",
embedding_dimension=384,
provider_id=qdrant_provider.provider_id,
)
```
Other than that, I just followed what was in
https://llama-stack.readthedocs.io/en/latest/getting_started/index.html
It would be good to have automated tests for this in the future, but
that would be a big undertaking.
Signed-off-by: Bill Murdock <bmurdock@redhat.com>
# What does this PR do?
Catches a bug in the previous codegen which was removing newlines.
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
```
python llama_stack/scripts/distro_codegen.py
```
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
# What does this PR do?
Catches docs up to source with:
```
python llama_stack/scripts/distro_codegen.py
```
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
[Describe the tests you ran to verify your changes with result
summaries. *Provide clear instructions so the plan can be easily
re-executed.*]
Manually checked
```
sphinx-autobuild docs/source build/html
```
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
# What does this PR do?
[Provide a short summary of what this PR does and why. Link to relevant
issues if applicable.]
Updated `uv.lock` to reflect the latest versions of `llama-models`,
`llama-stack`, and `llama-stack-client` (bumped to 0.1.2). This ensures
dependency consistency and avoids potential issues with outdated package
references.
Added `uv-sync` hook from `uv-pre-commit` repository to ensure
synchronization of dependencies.
Signed-off-by: Sébastien Han <seb@redhat.com>
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
[Describe the tests you ran to verify your changes with result
summaries. *Provide clear instructions so the plan can be easily
re-executed.*]
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
This PR removes the warnings when running tests for `remote::vllm`
provider:
```
Detected the chat template content format to be 'openai'. You can set `--chat-template-content-format` to override this.
```
## Test Plan
All tests passed without the warning messages shown above.
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
[Provide a short summary of what this PR does and why. Link to relevant
issues if applicable.]
Updated requirements.txt
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
[Describe the tests you ran to verify your changes with result
summaries. *Provide clear instructions so the plan can be easily
re-executed.*]
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
---------
Co-authored-by: sarthakdeshpande <sarthak.deshpande@engati.com>
# What does this PR do?
This significantly shortens the test time (about 10x faster) since most
of the time is spent on outputing the tokens "there are several planets
in our solar system that have...". We want to have an answer quicker,
especially when testing even larger models.
## Test Plan
```
LLAMA_STACK_BASE_URL=http://localhost:5002 pytest -v tests/client-sdk/inference/test_text_inference.py -k "test_text_chat_completion_non_streaming or test_text_chat_completion_streaming"
================================================================== test session starts ===================================================================
platform linux -- Python 3.10.16, pytest-8.3.4, pluggy-1.5.0 -- /home/yutang/.conda/envs/myenv/bin/python3.10
cachedir: .pytest_cache
rootdir: /home/yutang/repos/llama-stack
configfile: pyproject.toml
plugins: anyio-4.7.0
collected 12 items / 8 deselected / 4 selected
tests/client-sdk/inference/test_text_inference.py::test_text_chat_completion_non_streaming[meta-llama/Llama-3.1-8B-Instruct-Which planet do humans live on?-Earth] PASSED [ 25%]
tests/client-sdk/inference/test_text_inference.py::test_text_chat_completion_non_streaming[meta-llama/Llama-3.1-8B-Instruct-Which planet has rings around it with a name starting with letter S?-Saturn] PASSED [ 50%]
tests/client-sdk/inference/test_text_inference.py::test_text_chat_completion_streaming[meta-llama/Llama-3.1-8B-Instruct-What's the name of the Sun in latin?-Sol] PASSED [ 75%]
tests/client-sdk/inference/test_text_inference.py::test_text_chat_completion_streaming[meta-llama/Llama-3.1-8B-Instruct-What is the name of the US captial?-Washington] PASSED [100%]
```
---------
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
[Provide a short summary of what this PR does and why. Link to relevant
issues if applicable.]
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
[Describe the tests you ran to verify your changes with result
summaries. *Provide clear instructions so the plan can be easily
re-executed.*]
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
# What does this PR do?
Added examples (Section 4) of using Llama Stack 0.1 distro on together
and Llama 3.2 to answer questions about an image with LS Chat and Agent
APIs.
# What does this PR do?
Moved model availability check logic into a dedicated
check_model_availability function. Eliminated redundant code by reusing
the helper function in both embedding and non-embedding model
registration.
Signed-off-by: Sébastien Han <seb@redhat.com>
## Test Plan
Run Ollama and serve 2 models to get most the unit test pass:
```
ollama run llama3.2:3b-instruct-fp16 --keepalive 2m &
ollama run llama3.1:8b --keepalive 2m &
```
Run the unit test:
```
uv run pytest -v -k "ollama" --inference-model=llama3.2:3b-instruct-fp16 llama_stack/providers/tests/inference/test_model_registration.py
/Users/leseb/Documents/AI/llama-stack/.venv/lib/python3.13/site-packages/pytest_asyncio/plugin.py:207: PytestDeprecationWarning: The configuration option "asyncio_default_fixture_loop_scope" is unset.
The event loop scope for asynchronous fixtures will default to the fixture caching scope. Future versions of pytest-asyncio will default the loop scope for asynchronous fixtures to function scope. Set the default fixture loop scope explicitly in order to avoid unexpected behavior in the future. Valid fixture loop scopes are: "function", "class", "module", "package", "session"
warnings.warn(PytestDeprecationWarning(_DEFAULT_FIXTURE_LOOP_SCOPE_UNSET))
============================================ test session starts =============================================
platform darwin -- Python 3.13.1, pytest-8.3.4, pluggy-1.5.0 -- /Users/leseb/Documents/AI/llama-stack/.venv/bin/python3
cachedir: .pytest_cache
metadata: {'Python': '3.13.1', 'Platform': 'macOS-15.3-arm64-arm-64bit-Mach-O', 'Packages': {'pytest': '8.3.4', 'pluggy': '1.5.0'}, 'Plugins': {'html': '4.1.1', 'metadata': '3.1.1', 'asyncio': '0.25.3', 'anyio': '4.8.0', 'nbval': '0.11.0'}}
rootdir: /Users/leseb/Documents/AI/llama-stack
configfile: pyproject.toml
plugins: html-4.1.1, metadata-3.1.1, asyncio-0.25.3, anyio-4.8.0, nbval-0.11.0
asyncio: mode=Mode.STRICT, asyncio_default_fixture_loop_scope=None
collected 65 items / 60 deselected / 5 selected
llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_register_unsupported_model[-ollama] PASSED [ 20%]
llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_register_nonexistent_model[-ollama] PASSED [ 40%]
llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_register_with_llama_model[-ollama] FAILED [ 60%]
llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_initialize_model_during_registering[-ollama] FAILED [ 80%]
llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_register_with_invalid_llama_model[-ollama] PASSED [100%]
================================================== FAILURES ==================================================
_______________________ TestModelRegistration.test_register_with_llama_model[-ollama] ________________________
llama_stack/providers/tests/inference/test_model_registration.py:54: in test_register_with_llama_model
_ = await models_impl.register_model(
llama_stack/providers/utils/telemetry/trace_protocol.py:91: in async_wrapper
result = await method(self, *args, **kwargs)
llama_stack/distribution/routers/routing_tables.py:245: in register_model
registered_model = await self.register_object(model)
llama_stack/distribution/routers/routing_tables.py:192: in register_object
registered_obj = await register_object_with_provider(obj, p)
llama_stack/distribution/routers/routing_tables.py:53: in register_object_with_provider
return await p.register_model(obj)
llama_stack/providers/utils/telemetry/trace_protocol.py:91: in async_wrapper
result = await method(self, *args, **kwargs)
llama_stack/providers/remote/inference/ollama/ollama.py:368: in register_model
await check_model_availability(model.provider_resource_id)
llama_stack/providers/remote/inference/ollama/ollama.py:359: in check_model_availability
raise ValueError(
E ValueError: Model 'custom-model' is not available in Ollama. Available models: llama3.1:8b, llama3.2:3b-instruct-fp16
__________________ TestModelRegistration.test_initialize_model_during_registering[-ollama] ___________________
llama_stack/providers/tests/inference/test_model_registration.py:85: in test_initialize_model_during_registering
mock_load_model.assert_called_once()
/opt/homebrew/Cellar/python@3.13/3.13.1/Frameworks/Python.framework/Versions/3.13/lib/python3.13/unittest/mock.py:956: in assert_called_once
raise AssertionError(msg)
E AssertionError: Expected 'load_model' to have been called once. Called 0 times.
-------------------------------------------- Captured stderr call --------------------------------------------
W0207 11:55:26.777000 90854 .venv/lib/python3.13/site-packages/torch/distributed/elastic/multiprocessing/redirects.py:29] NOTE: Redirects are currently not supported in Windows or MacOs.
========================================== short test summary info ===========================================
FAILED llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_register_with_llama_model[-ollama] - ValueError: Model 'custom-model' is not available in Ollama. Available models: llama3.1:8b, llama3.2:3b-i...
FAILED llama_stack/providers/tests/inference/test_model_registration.py::TestModelRegistration::test_initialize_model_during_registering[-ollama] - AssertionError: Expected 'load_model' to have been called once. Called 0 times.
=========================== 2 failed, 3 passed, 60 deselected, 2 warnings in 1.84s ===========================
```
We only "care" about the `test_register_nonexistent_model` for this
code.
## Sources
Please link relevant resources if necessary.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
currently this is the output when you run a distribution locally without
running `llama stack build`:
```
Traceback (most recent call last):
File "/Users/charliedoern/Documents/llama-sdk.py", line 25, in <module>
models = client.models.list()
^^^^^^^^^^^^^^^^^^^^
File "/Users/charliedoern/Documents/llama-stack-client-python/src/llama_stack_client/resources/models.py", line 107, in list
raise exc
File "/Users/charliedoern/Documents/llama-stack-client-python/src/llama_stack_client/resources/models.py", line 95, in list
return self._get(
^^^^^^^^^^
File "/Users/charliedoern/Documents/llama-stack-client-python/src/llama_stack_client/_base_client.py", line 1212, in get
return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/charliedoern/Documents/llama-stack/llama_stack/distribution/library_client.py", line 168, in request
return asyncio.run(self.async_client.request(*args, **kwargs))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/homebrew/Cellar/python@3.11/3.11.10/Frameworks/Python.framework/Versions/3.11/lib/python3.11/asyncio/runners.py", line 190, in run
return runner.run(main)
^^^^^^^^^^^^^^^^
File "/opt/homebrew/Cellar/python@3.11/3.11.10/Frameworks/Python.framework/Versions/3.11/lib/python3.11/asyncio/runners.py", line 118, in run
return self._loop.run_until_complete(task)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/homebrew/Cellar/python@3.11/3.11.10/Frameworks/Python.framework/Versions/3.11/lib/python3.11/asyncio/base_events.py", line 654, in run_until_complete
return future.result()
^^^^^^^^^^^^^^^
File "/Users/charliedoern/Documents/llama-stack/llama_stack/distribution/library_client.py", line 258, in request
if not self.endpoint_impls:
^^^^^^^^^^^^^^^^^^^
AttributeError: 'AsyncLlamaStackAsLibraryClient' object has no attribute 'endpoint_impls'
```
the intended exception is never raised, add an except for an
AttributeError so users can catch when they call things like
`models.list()` and so that a more useful error telling them that the
client is not properly initialized is printed.
## Test Plan
Please describe:
- I ran the script found here:
https://llama-stack.readthedocs.io/en/latest/getting_started/index.html#run-inference-with-python-sdk
locally with the changes in this PR and the exception was caught
successfully.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
---------
Signed-off-by: Charlie Doern <cdoern@redhat.com>
Co-authored-by: Ashwin Bharambe <ashwin.bharambe@gmail.com>
# What does this PR do?
Previously, the test was failing due to a pydantic validation error
caused by passing raw binary image data instead of a valid Unicode
string. This fix encodes the image data as base64, ensuring it is a
valid string format compatible with `ImageContentItem`.
Error:
```
______________ ERROR collecting llama_stack/providers/tests/inference/test_vision_inference.py _______________
llama_stack/providers/tests/inference/test_vision_inference.py:31: in <module>
class TestVisionModelInference:
llama_stack/providers/tests/inference/test_vision_inference.py:37: in TestVisionModelInference
ImageContentItem(image=dict(data=PASTA_IMAGE)),
E pydantic_core._pydantic_core.ValidationError: 1 validation error for ImageContentItem
E image.data
E Input should be a valid string, unable to parse raw data as a unicode string [type=string_unicode, input_value=b'\xff\xd8\xff\xe0\x00\x1...0\xe6\x9f5\xb5?\xff\xd9', input_type=bytes]
E For further information visit
https://errors.pydantic.dev/2.10/v/string_unicode
```
Signed-off-by: Sébastien Han <seb@redhat.com>
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
Execute the following:
```
ollama run llama3.2-vision --keepalive 2m &
uv run pytest -v -s -k "ollama" --inference-model=llama3.2-vision:latest llama_stack/providers/tests/inference/test_vision_inference.py
llama_stack/providers/tests/inference/test_vision_inference.py::TestVisionModelInference::test_vision_chat_completion_non_streaming[-ollama-image0-expected_strings0] PASSED
llama_stack/providers/tests/inference/test_vision_inference.py::TestVisionModelInference::test_vision_chat_completion_non_streaming[-ollama-image1-expected_strings1] FAILED
llama_stack/providers/tests/inference/test_vision_inference.py::TestVisionModelInference::test_vision_chat_completion_streaming[-ollama] FAILED
```
The last two tests are failing because Cloudflare blocked me from
accessing
https://www.healthypawspetinsurance.com/Images/V3/DogAndPuppyInsurance/Dog_CTA_Desktop_HeroImage.jpg
but this has no impact on the current fix.
[//]: # (## Documentation)
[//]: # (- [ ] Added a Changelog entry if the change is significant)
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
Enables HTTPS option for Llama Stack.
While doing so, introduces a `ServerConfig` sub-structure to house all
server related configuration (port, ssl, etc.)
Also simplified the `start_container.sh` entrypoint to simply be
`python` instead of a complex bash command line.
## Test Plan
Conda:
Run:
```bash
$ llama stack build --template together
$ llama stack run --port 8322 # ensure server starts
$ llama-stack-client configure --endpoint http://localhost:8322
$ llama-stack-client models list
```
Create a self-signed SSL key / cert pair. Then, using a local checkout
of `llama-stack-client-python`, change
https://github.com/meta-llama/llama-stack-client-python/blob/main/src/llama_stack_client/_base_client.py#L759
to add `kwargs.setdefault("verify", False)` so SSL verification is
disabled. Then:
```bash
$ llama stack run --port 8322 --tls-keyfile <KEYFILE> --tls-certfile <CERTFILE>
$ llama-stack-client configure --endpoint https://localhost:8322 # notice the `https`
$ llama-stack-client models list
```
Also tested with containers (but of course one needs to make sure the
cert and key files are appropriately provided to the container.)
# What does this PR do?
This PR splits the inference tests into text and vision to make testing
on vLLM provider easier as mentioned in
https://github.com/meta-llama/llama-stack/pull/951 since serving
multiple models (e.g. Llama-3.2-11B-Vision-Instruct and
Llama-3.1-8B-Instruct) on a single port using the OpenAI API is [not
supported yet](https://docs.vllm.ai/en/v0.5.5/serving/faq.html) so it's
a bit tricky to test both at the same time.
## Test Plan
All previously passing tests related to text still pass:
`LLAMA_STACK_BASE_URL=http://localhost:5002 pytest -v
tests/client-sdk/inference/test_text_inference.py`
All vision tests passed via `LLAMA_STACK_BASE_URL=http://localhost:5002
pytest -v tests/client-sdk/inference/test_vision_inference.py`.
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
Imported `ToolConfig` from the `llama_stack.apis.inference` module to
resolve missing reference and ensure proper functionality within the
`groq.py` file.
Signed-off-by: Sébastien Han <seb@redhat.com>
## Test Plan
Without the change, pytest will run with the following error:
```
uv run pytest -v -s -k "ollama" llama_stack/providers/tests/
/Users/leseb/Documents/AI/llama-stack/.venv/lib/python3.13/site-packages/pytest_asyncio/plugin.py:207: PytestDeprecationWarning: The configuration option "asyncio_default_fixture_loop_scope" is unset.
The event loop scope for asynchronous fixtures will default to the fixture caching scope. Future versions of pytest-asyncio will default the loop scope for asynchronous fixtures to function scope. Set the default fixture loop scope explicitly in order to avoid unexpected behavior in the future. Valid fixture loop scopes are: "function", "class", "module", "package", "session"
warnings.warn(PytestDeprecationWarning(_DEFAULT_FIXTURE_LOOP_SCOPE_UNSET))
============================================ test session starts =============================================
platform darwin -- Python 3.13.1, pytest-8.3.4, pluggy-1.5.0 -- /Users/leseb/Documents/AI/llama-stack/.venv/bin/python3
cachedir: .pytest_cache
metadata: {'Python': '3.13.1', 'Platform': 'macOS-15.3-arm64-arm-64bit-Mach-O', 'Packages': {'pytest': '8.3.4', 'pluggy': '1.5.0'}, 'Plugins': {'html': '4.1.1', 'metadata': '3.1.1', 'asyncio': '0.25.3', 'anyio': '4.8.0', 'nbval': '0.11.0'}}
rootdir: /Users/leseb/Documents/AI/llama-stack
configfile: pyproject.toml
plugins: html-4.1.1, metadata-3.1.1, asyncio-0.25.3, anyio-4.8.0, nbval-0.11.0
asyncio: mode=Mode.STRICT, asyncio_default_fixture_loop_scope=None
collected 379 items / 1 error / 349 deselected / 30 selected
=================================================== ERRORS ===================================================
__________________ ERROR collecting llama_stack/providers/tests/inference/groq/test_init.py __________________
llama_stack/providers/tests/inference/groq/test_init.py:11: in <module>
from llama_stack.providers.remote.inference.groq.groq import GroqInferenceAdapter
llama_stack/providers/remote/inference/groq/groq.py:72: in <module>
class GroqInferenceAdapter(Inference, ModelRegistryHelper, NeedsRequestProviderData):
llama_stack/providers/remote/inference/groq/groq.py:102: in GroqInferenceAdapter
tool_config: Optional[ToolConfig] = None,
E NameError: name 'ToolConfig' is not defined
========================================== short test summary info ===========================================
ERROR llama_stack/providers/tests/inference/groq/test_init.py - NameError: name 'ToolConfig' is not defined
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Interrupted: 1 error during collection !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
=============================== 349 deselected, 22 warnings, 1 error in 0.28s ================================
```
With the change the test continues to run and fails with a different
error:
```
uv run pytest -v -s llama_stack/providers/tests/
/Users/leseb/Documents/AI/llama-stack/.venv/lib/python3.13/site-packages/pytest_asyncio/plugin.py:207: PytestDeprecationWarning: The configuration option "asyncio_default_fixture_loop_scope" is unset.
The event loop scope for asynchronous fixtures will default to the fixture caching scope. Future versions of pytest-asyncio will default the loop scope for asynchronous fixtures to function scope. Set the default fixture loop scope explicitly in order to avoid unexpected behavior in the future. Valid fixture loop scopes are: "function", "class", "module", "package", "session"
warnings.warn(PytestDeprecationWarning(_DEFAULT_FIXTURE_LOOP_SCOPE_UNSET))
============================================ test session starts =============================================
platform darwin -- Python 3.13.1, pytest-8.3.4, pluggy-1.5.0 -- /Users/leseb/Documents/AI/llama-stack/.venv/bin/python3
cachedir: .pytest_cache
metadata: {'Python': '3.13.1', 'Platform': 'macOS-15.3-arm64-arm-64bit-Mach-O', 'Packages': {'pytest': '8.3.4', 'pluggy': '1.5.0'}, 'Plugins': {'html': '4.1.1', 'metadata': '3.1.1', 'asyncio': '0.25.3', 'anyio': '4.8.0', 'nbval': '0.11.0'}}
rootdir: /Users/leseb/Documents/AI/llama-stack
configfile: pyproject.toml
plugins: html-4.1.1, metadata-3.1.1, asyncio-0.25.3, anyio-4.8.0, nbval-0.11.0
asyncio: mode=Mode.STRICT, asyncio_default_fixture_loop_scope=None
collected 342 items / 1 error
=================================================== ERRORS ===================================================
______________ ERROR collecting llama_stack/providers/tests/inference/test_vision_inference.py _______________
llama_stack/providers/tests/inference/test_vision_inference.py:29: in <module>
class TestVisionModelInference:
llama_stack/providers/tests/inference/test_vision_inference.py:35: in TestVisionModelInference
ImageContentItem(image=dict(data=PASTA_IMAGE)),
E pydantic_core._pydantic_core.ValidationError: 1 validation error for ImageContentItem
E image.data
E Input should be a valid string, unable to parse raw data as a unicode string [type=string_unicode, input_value=b'\xff\xd8\xff\xe0\x00\x1...0\xe6\x9f5\xb5?\xff\xd9', input_type=bytes]
E For further information visit https://errors.pydantic.dev/2.10/v/string_unicode
========================================== short test summary info ===========================================
ERROR llama_stack/providers/tests/inference/test_vision_inference.py - pydantic_core._pydantic_core.ValidationError: 1 validation error for ImageContentItem
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Interrupted: 1 error during collection !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
======================================= 22 warnings, 1 error in 0.25s ========================================
```
Which is fixed in https://github.com/meta-llama/llama-stack/pull/1003.
## Sources
Please link relevant resources if necessary.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Sébastien Han <seb@redhat.com>
We use weekly releases as a way to communicate important improvements.
Keeping this information synced across is more overhead than we have
bandwidth for right now. We may change this process over time.
Fixes#966.
Verified that:
1. Correct list of APIs are printed out when running `llama stack
list-providers`
2. `llama stack list-providers <api>` works as expected.
---------
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
Refactored tests by removing unused exception alias (as exc_info) in
pytest.raises, improving code clarity and reducing lint warnings.
exc_info was never used.
Signed-off-by: Sébastien Han <seb@redhat.com>
## Test Plan
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
## Sources
Please link relevant resources if necessary.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
Fix link
## Test Plan
<!--
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
-->
<!--
## Sources
Please link relevant resources if necessary.
-->
<!--
## Documentation
- [ ] Added a
[Changelog](https://github.com/meta-llama/llama-stack/blob/main/CHANGELOG.md)
entry if the change is significant (new feature, breaking change etc.).
-->
# What does this PR do?
In several examples we use the same faiss index , which means running it
multiple times fills up the index with duplicates which eventually
degrades the model performance on RAG as multiple copies of the same
irrelevant chunks might be picked up several times.
Fix is to ensure we create a new index each time.
Resolves issue in this discussion -
https://github.com/meta-llama/llama-stack/discussions/995
## Test Plan
Re-ran the getting started guide multiple times to see the same output
Co-authored-by: Hardik Shah <hjshah@fb.com>
# What does this PR do?
<!-- Provide a short summary of what this PR does and why. Usually, the
relevant context should be present in a linked issue. -->
Corrects some typographical errors found in the
`docs/zero_to_hero_guide/README.md` file.
<!-- Uncomment this section with the issue number if an issue is being
resolved
**Issue resolved by this Pull Request:** Closes #
--->
## Test Plan
<!--
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
-->
N/A
<!--
## Sources
Please link relevant resources if necessary.
-->
<!--
## Documentation
- [ ] Added a
[Changelog](https://github.com/meta-llama/llama-stack/blob/main/CHANGELOG.md)
entry if the change is significant (new feature, breaking change etc.).
-->
Co-authored-by: Maxime Lecanu <mlecanu@fb.com>
This adds a new workflow to check semantic PR titles to match the
[Conventional Commits spec](https://www.conventionalcommits.org/). This
will make it easier to browse commit history and enable automation in
the future.
---------
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
This PR moves some content from [the recent blog
post](https://blog.vllm.ai/2025/01/27/intro-to-llama-stack-with-vllm.html)
to here as a more official guide for users who'd like to deploy Llama
Stack on Kubernetes.
---------
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
Switch to uv for dependency management and update CONTRIBUTING.md with
new setup instructions. Add missing dev dependencies to pyproject.toml
and apply minor formatting fixes.
Signed-off-by: Sébastien Han <seb@redhat.com>
- [ ] Addresses issue (#issue)
## Test Plan
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
## Sources
Please link relevant resources if necessary.
## Before submitting
- [x] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
Replaced references to `memory` with `vector_io` in
`DEFAULT_PROVIDER_COMBINATIONS` and adjusted corresponding fixture
imports to ensure proper configuration for vector I/O during tests. This
change aligns with the new testing structure.
Followup of https://github.com/meta-llama/llama-stack/pull/830 when the
memory fixture was removed.
Signed-off-by: Sébastien Han <seb@redhat.com>
## Test Plan
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
## Sources
Please link relevant resources if necessary.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
very small fix I noticed some unused arguments, but this seems like the
easiest one to remove since its passed in explicitly.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Charlie Doern <cdoern@redhat.com>
Also, hiding guidance to the author under comments to avoid polluting
the description with ti.
Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
# What does this PR do?
Using `Closes #` syntax in PR template, as per:
https://docs.github.com/en/get-started/writing-on-github/working-with-advanced-formatting/using-keywords-in-issues-and-pull-requests
```
In short, provide a summary of what this PR does and why. Usually, the relevant context should be present in a linked issue.
```
Hides this ^.
```
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
```
And this ^.
```
Please link relevant resources if necessary.
```
And this ^.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [x] Ran pre-commit to handle lint / formatting issues.
- [x] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
# What does this PR do?
- Added a checklist item in the PR template to ensure significant
changes are documented in the changelog.
- Updated `CHANGELOG.md` with a placeholder for version `0.2.0`.
- This is an effort to resurrect the consistent usage of the changelog
file.
Signed-off-by: Sébastien Han <seb@redhat.com>
## Test Plan
Please describe:
- tests you ran to verify your changes with result summaries.
- provide instructions so it can be reproduced.
## Sources
Please link relevant resources if necessary.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
The current default system prompt for llama3.2 tends to overindex on
tool calling and doesn't work well when the prompt does not require tool
calling.
This PR adds an option to override the default system prompt, and
organizes tool-related configs into a new config object.
- [ ] Addresses issue (#issue)
## Test Plan
LLAMA_STACK_CONFIG=together pytest
\-\-inference\-model=meta\-llama/Llama\-3\.3\-70B\-Instruct -s -v
tests/client-sdk/agents/test_agents.py::test_override_system_message_behavior
## Sources
Please link relevant resources if necessary.
## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Ran pre-commit to handle lint / formatting issues.
- [ ] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
Pull Request section?
- [ ] Updated relevant documentation.
- [ ] Wrote necessary unit or integration tests.
