# What does this PR do?
Closes#1968.
The asynchronous client in `VLLMInferenceAdapter` is now initialized
directly before first use and not in `VLLMInferenceAdapter.initialize`.
This prevents issues arising due to accessing an expired event loop from
a completed `asyncio.run`.
## Test Plan
Ran unit tests, including `test_remote_vllm.py`.
Ran the code snippet mentioned in #1968.
---------
Co-authored-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
Now, tool outputs and retrieved chunks from the vector DB (i.e.,
everything except for the actual model reply) are hidden under an
expander form when presented to the user.
# Test Plan
Navigate to the RAG page in the Playground UI.
# What does this PR do?
The together inference provider was throwing a stack trace every time it
shut down, as it was trying to call a non-existent `close` method on the
AsyncTogether client. While fixing that, I also adjusted its shutdown
logic to close the OpenAI client if we've created one of those, as that
client does have a `close` method.
In testing that, I also realized we were defaulting to treating all
requests as streaming requests instead of defaulting to non-streaming.
So, this flips that default to non-streaming to match how the other
providers work.
## Test Plan
I tested this by ensuring the together inference provider no longer
spits out a long stack trace when shutting it down and by running the
OpenAI API chat completion verification suite to ensure the change in
default streaming logic didn't mess anything else up.
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
This PR cleans up the sidebar on the tools page of the playground in the
following ways:
* created a clearer hierarchy of configuration options and tool
selections.
* Removed the `mcp::` or `builtin::` prefixes from the tool selection
buttons.
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
Run the playground and see the updated sidebar does not cause any new
errors.
```
streamlit run llama_stack/distribution/ui/app.py
```
[//]: # (## Documentation)
Signed-off-by: Michael Clifford <mcliffor@redhat.com>
37da47ef8e (diff-4d7c51b1efe9043e44439a949dfd92e5827321b34082903477fd04876edb7552)
Pydantic was updated from v1 to v2 in this commit which caused this
breaking change
# What does this PR do?
Part of #1857
This won't fix the Validation error with the example, but it will
correctly supply user with a proper error rather than a 5xx code.
Signed-off-by: Kevin <kpostlet@redhat.com>
# What does this PR do?
We were passing a dict into the compat mixin for OpenAI Completions when
using Llama models with Fireworks, and that was breaking some strong
typing code that was added in openai_compat.py. We shouldn't have been
converting these params to a dict in that case anyway, so this adjusts
things to pass the params in as their actual original types when calling
the OpenAIChatCompletionToLlamaStackMixin.
## Test Plan
All of the fireworks provider verification tests were failing due to
some OpenAI compatibility cleanup in #1962. The changes in that PR were
good to make, and this just cleans up the fireworks provider code to
stop passing in untyped dicts to some of those `openai_compat.py`
methods since we have the original strongly-typed parameters we can pass
in.
```
llama stack run --image-type venv tests/verifications/openai-api-verification-run.yaml
```
```
python -m pytest -s -v tests/verifications/openai_api/test_chat_completion.py --provider=fireworks-llama-stack
```
Before this PR, all of the fireworks OpenAI verification tests were
failing. Now, most of them are passing.
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
- Update NVIDIA documentation links to GA docs
- Remove reference to notebooks until merged
[//]: # (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)
Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
# What does this PR do?
This is helpful when debugging issues with vLLM + Llama Stack after this
PR https://github.com/vllm-project/vllm/pull/15593
---------
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
# What does this PR do?
NVIDIA Inference provider was using the ModelRegistryHelper to map input
model ids to provider model ids. this updates it to use the model_store.
## Test Plan
`LLAMA_STACK_CONFIG=http://localhost:8321 uv run pytest -v
tests/integration/inference/{test_embedding.py,test_text_inference.py,test_openai_completion.py}
--embedding-model nvidia/llama-3.2-nv-embedqa-1b-v2
--text-model=meta-llama/Llama-3.1-70B-Instruct`
# What does this PR do?
Fixes the UBI 9 container build failure ( `error: command 'gcc' failed`
when installing `polyleven`, `faiss`, etc.) by installing the missing
compiler tool‑chain:
- `python3.11-devel gcc` make added to the UBI 9 `dnf install` line.
### Closes#1970
## Test Plan
- Build a distro with an UBI image
Test plan:
python tests/verifications/generate_report.py --providers
fireworks,together,llama_meta_ref,openai
Co-authored-by: Eric Huang <erichuang@fb.com>
# What does this PR do?
Allow users to name an agent and use the name in telemetry instead of
relying on randomly generated agent_ids. This improves the developer
experience by making it easier to find specific agents in telemetry
logs.
Closes#1832
## Test Plan
- Added tests to verify the agent name is properly stored and retrieved
- Ran `uv run -- pytest -v
tests/integration/telemetry/test_telemetry.py::test_agent_name_filtering`
from the root of the project and made sure the tests pass
- Ran `uv run -- pytest -v
tests/integration/telemetry/test_telemetry.py::test_agent_query_spans`
to verify existing code without agent names still works correctly
## Use Example
```
agent = Agent(
llama_stack_client,
model=text_model_id,
name="CustomerSupportAgent", # New parameter
instructions="You are a helpful customer support assistant"
)
session_id = agent.create_session(f"test-session-{uuid4()}")
```
## Implementation Notes
- Agent names are optional string parameters with no additional
validation
- Names are not required to be unique - multiple agents can have the
same name
- The agent_id remains the unique identifier for an agent
---------
Co-authored-by: raghotham <raghotham@gmail.com>
# What does this PR do?
Some of our multi-turn verification tests were failing because I had
accidentally marked content as a required field in the OpenAI chat
completion request assistant messages, but it's actually optional. It is
required for messages from other roles, but assistant is explicitly
allowed to be optional.
Similarly, the assistant message tool_calls field should default to None
instead of an empty list.
These two changes get the openai-llama-stack verification test back to
100% passing, just like it passes 100% when not behind Llama Stack. They
also increase the pass rate of some of the other providers in the
verification test, but don't get them to 100%.
## Test Plan
I started a Llama Stack server setup to run all the verification tests
(requires OPENAI_API_KEY env variable)
```
llama stack run --image-type venv tests/verifications/openai-api-verification-run.yaml
```
Then, I manually ran the verification tests to see which were failing,
fix them, and ran them again after these changes to ensure they were all
passing.
```
python -m pytest -s -v tests/verifications/openai_api/test_chat_completion.py --provider=openai-llama-stack
```
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
Add NVIDIA platform docs that serve as a starting point for Llama Stack
users and explains all supported microservices.
[//]: # (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)
---------
Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
# What does this PR do?
This PR handles the case where a Customization Job's status is
`unknown`. Since we don't map `unknown` to a valid `JobStatus`, the
PostTraining provider throws an exception when fetching/listing a job.
[//]: # (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.*]
`./scripts/unit-tests.sh
tests/unit/providers/nvidia/test_supervised_fine_tuning.py` succeeds
[//]: # (## Documentation)
Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
# What does this PR do?
Fixes a crash that occurred when building a stack as a container image
via the interactive wizard without supplying --template or --config.
- Root cause: template_or_config was None; only the container path
relies on that parameter, which later reaches subprocess.run() and
triggers
`TypeError: expected str, bytes or os.PathLike object, not NoneType.`
- Change: in `_run_stack_build_command_from_build_config` we now fall
back to the freshly‑written build‑spec file whenever both optional
sources are missing. Also adds a spy‑based unit test that asserts a
valid string path is passed to build_image() for container builds.
### Closes#1976
## Test Plan
- New unit test: test_build_path.py. Monkey‑patches build_image,
captures the fourth argument, and verifies it is a real path
- Manual smoke test:
```
llama stack build --image-type container
# answer wizard prompts
```
Build proceeds into Docker without raising the previous TypeError.
## Future Work
Harmonise `build_image` arguments so every image type receives the same
inputs, eliminating this asymmetric special‑case.
# What does this PR do?
Build failures are hard to read, sometimes we get errors like:
```
Error building stack: 'key'
```
Which are difficult to debug without a proper trace.
## Test Plan
If `llama stack build` fails you get a traceback now.
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
This PR lets users select an existing vdb to use with their agent on the
tools page of the playground. The drop down menu that lets users select
a vdb only appears when the rag tool is selected. Without this change,
there is no way for a user to specify which vdb they want their rag tool
to use on the tools page. I have intentionally left the RAG options
sparse here since the full RAG options are exposed on the RAG page.
## Test Plan
Without these changes the RAG tool will throw the following error:
`name: knowledge_search) does not have any content `
With these changes the RAG tool works as expected.
Signed-off-by: Michael Clifford <mcliffor@redhat.com>
# What does this PR do?
Adds `meta/llama-3.2-1b-instruct` to list of models that NeMo Customizer
can fine-tune. This is the model our example notebooks typically use for
fine-tuning.
[//]: # (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)
Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
Fixes: #1955
Since 0.2.0, the vLLM gets an empty list (vs ``None``in 0.1.9 and
before) when there are no tools configured which causes the issue
described in #1955 p. This patch avoids sending the 'tools' param to the
vLLM altogether instead of an empty list.
It also adds a small unit test to avoid regressions.
The OpenAI
[specification](https://platform.openai.com/docs/api-reference/chat/create)
does not explicitly state that the list cannot be empty but I found this
out through experimentation and it might depend on the actual remote
vllm. In any case, as this parameter is Optional, is best to skip it
altogether if there's no tools configured.
Signed-off-by: Daniel Alvarez <dalvarez@redhat.com>
# What does this PR do?
This PR adds a `max_tokens` slider to playground tools page. I have
found that in some instances the llama stack server throws a 500 error
if the max_tokens value is not explicitly set in the agent's
`sampling_params`. This PR, uses the same implementation of the
`max_tokens` slider from the chat page, and includes it on the tools
page.
## Test Plan
1. Attempting to call a tool without these changes results in a `500:
Internal server error: An unexpected error occurred`.
2. Attempting to call a tool with these changes results in the expected
output.
Signed-off-by: Michael Clifford <mcliffor@redhat.com>
# What does this PR do?
PR adds instructions to setup vLLM remote endpoint for vllm-remote llama
stack distribution.
## Test Plan
* Verified with manual tests of the configured vllm-remote against vllm
endpoint running on the system with Intel GPU
* Also verified with ci pytests (see cmdline below). Test passes in the
same capacity as it does on the A10 Nvidia setup (some tests do fail
which seems to be known issues with vllm remote llama stack
distribution)
```
pytest -s -v tests/integration/inference/test_text_inference.py \
--stack-config=http://localhost:5001 \
--text-model=meta-llama/Llama-3.2-3B-Instruct
```
CC: @ashwinb
Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
# What does this PR do?
allow users to specify only the providers they want in the llama stack
build command. If a user wants a non-interactive build, but doesn't want
to use a template, `--providers` allows someone to specify something
like `--providers inference=remote::ollama` for a distro with JUST
ollama
## Test Plan
`llama stack build --providers inference=remote::ollama --image-type
venv`
<img width="1084" alt="Screenshot 2025-03-20 at 9 34 14 AM"
src="https://github.com/user-attachments/assets/502b5fa2-edab-4267-a595-4f987204a6a9"
/>
`llama stack run --image-type venv
/Users/charliedoern/projects/Documents/llama-stack/venv-run.yaml`
<img width="1149" alt="Screenshot 2025-03-20 at 9 35 19 AM"
src="https://github.com/user-attachments/assets/433765f3-6b7f-4383-9241-dad085b69228"
/>
---------
Signed-off-by: Charlie Doern <cdoern@redhat.com>
Signed-off-by: Sébastien Han <seb@redhat.com>
Co-authored-by: Sébastien Han <seb@redhat.com>
## What does this PR do?
This PR improves the server's request routing logic by ensuring built-in
FastAPI paths such as `/docs`, `/redoc`, `/openapi.json`,
`/favicon.ico`, and `/static` bypass the custom `TracingMiddleware`.
This prevents unnecessary tracing logic for documentation and static
file requests, ensuring better performance and cleaner logs.
Additionally, it adds proper metadata (`title`, `description`, and
`version`) to the FastAPI application initialization and updates the
requirements document accordingly.
[//]: # (Closes#1822 )
---
## Test Plan
- Ran the server locally with `uvicorn` using the provided `run.yaml`
config
- Verified that:
- FastAPI docs (`/docs`, `/redoc`) load correctly without triggering the
custom tracing middleware
- All other routes still go through the middleware and trace logic
- Application metadata appears as expected in the OpenAPI docs
To reproduce:
1. Start the server with `python server.py --template <template-name>`
2. Navigate to `/docs` and `/redoc`
3. Confirm that no extra trace headers are added for those routes
4. Confirm other API endpoints behave as expected and include
`x-trace-id` in the response headers
[//]: # (## Documentation)
---
Froze the requirements file to include many of the other libraries that
have been added in the past few releases to make install easier.
---------
Co-authored-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
ollama's CLI supports running models via commands such as 'ollama run
llama3.2' this syntax does not work with the INFERENCE_MODEL llamastack
var as currently specifying a tag such as 'latest' is required
this commit will check to see if the 'latest' model is available and use
that model if a user passes a model name without a tag but the 'latest'
is available in ollama
## Test Plan
Behavior pre-code change
```bash
$ INFERENCE_MODEL=llama3.2 llama stack build --template ollama --image-type venv --run
...
INFO 2025-04-08 13:42:42,842 llama_stack.providers.remote.inference.ollama.ollama:80 inference: checking
connectivity to Ollama at `http://beanlab1.bss.redhat.com:11434`...
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/server/server.py", line 502, in <module>
main()
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/server/server.py", line 401, in main
impls = asyncio.run(construct_stack(config))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib64/python3.12/asyncio/runners.py", line 195, in run
return runner.run(main)
^^^^^^^^^^^^^^^^
File "/usr/lib64/python3.12/asyncio/runners.py", line 118, in run
return self._loop.run_until_complete(task)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib64/python3.12/asyncio/base_events.py", line 691, in run_until_complete
return future.result()
^^^^^^^^^^^^^^^
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/stack.py", line 222, in construct_stack
await register_resources(run_config, impls)
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/stack.py", line 99, in register_resources
await method(**obj.model_dump())
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/providers/utils/telemetry/trace_protocol.py", line 102, in async_wrapper
result = await method(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/routers/routing_tables.py", line 294, in register_model
registered_model = await self.register_object(model)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/routers/routing_tables.py", line 228, in register_object
registered_obj = await register_object_with_provider(obj, p)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/distribution/routers/routing_tables.py", line 77, in register_object_with_provider
return await p.register_model(obj)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/providers/utils/telemetry/trace_protocol.py", line 102, in async_wrapper
result = await method(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/nathan/ai/llama-stack/repos/llama-stack/llama_stack/providers/remote/inference/ollama/ollama.py", line 315, in register_model
raise ValueError(
ValueError: Model 'llama3.2' is not available in Ollama. Available models: llama3.2:latest
++ error_handler 108
++ echo 'Error occurred in script at line: 108'
Error occurred in script at line: 108
++ exit 1
```
Behavior post-code change
```bash
$ INFERENCE_MODEL=llama3.2 llama stack build --template ollama --image-type venv --run
...
INFO 2025-04-08 13:58:17,365 llama_stack.providers.remote.inference.ollama.ollama:80 inference: checking
connectivity to Ollama at `http://beanlab1.bss.redhat.com:11434`...
WARNING 2025-04-08 13:58:18,190 llama_stack.providers.remote.inference.ollama.ollama:317 inference: Imprecise provider
resource id was used but 'latest' is available in Ollama - using 'llama3.2:latest'
INFO 2025-04-08 13:58:18,191 llama_stack.providers.remote.inference.ollama.ollama:308 inference: Pulling embedding
model `all-minilm:latest` if necessary...
INFO 2025-04-08 13:58:18,799 __main__:478 server: Listening on ['::', '0.0.0.0']:8321
INFO: Started server process [28378]
INFO: Waiting for application startup.
INFO 2025-04-08 13:58:18,803 __main__:148 server: Starting up
INFO: Application startup complete.
INFO: Uvicorn running on http://['::', '0.0.0.0']:8321 (Press CTRL+C to quit)
...
```
## Documentation
Did not document this anywhere but happy to do so if there is an
appropriate place
Signed-off-by: Nathan Weinberg <nweinber@redhat.com>
# What does this PR do?
Now a separate thread is started to execute training jobs. Training
requests now return job ID before the job completes. (Which fixes API
timeouts for any jobs that take longer than a minute.)
Note: the scheduler code is meant to be spun out in the future into a
common provider service that can be reused for different APIs and
providers. It is also expected to back the /jobs API proposed here:
https://github.com/meta-llama/llama-stack/discussions/1238
Hence its somewhat generalized form which is expected to simplify its
adoption elsewhere in the future.
Note: this patch doesn't attempt to implement missing APIs (e.g. cancel
or job removal). This work will belong to follow-up PRs.
[//]: # (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.*]
Added unit tests for the scheduler module. For the API coverage, did
manual testing and was able to run a training cycle on GPU. The initial
call returned job ID before the training completed, as (now) expected.
Artifacts are returned as expected.
```
JobArtifactsResponse(checkpoints=[{'identifier': 'meta-llama/Llama-3.2-3B-Instruct-sft-0', 'created_at': '2025-03-07T22:45:19.892714', 'epoch': 0, 'post_training_job_id': 'test-job2ee77104-2fd3-4a4e-84cf-f83f8b8f1f50', 'path': '/home/ec2-user/.llama/checkpoints/meta-llama/Llama-3.2-3B-Instruct-sft-0', 'training_metrics': None}], job_uuid='test-job2ee77104-2fd3-4a4e-84cf-f83f8b8f1f50')
```
The integration test is currently disabled for the provider. I will look
into how it can be enabled in a different PR / issue context.
[//]: # (## Documentation)
Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
# What does this PR do?
TLDR: Changes needed to get 100% passing tests for OpenAI API
verification tests when run against Llama Stack with the `together`,
`fireworks`, and `openai` providers. And `groq` is better than before,
at 88% passing.
This cleans up the OpenAI API support for image message types
(specifically `image_url` types) and handling of the `response_format`
chat completion parameter. Both of these required a few more Pydantic
model definitions in our Inference API, just to move from the
not-quite-right stubs I had in place to something fleshed out to match
the actual OpenAI API specs.
As part of testing this, I also found and fixed a bug in the litellm
implementation of openai_completion and openai_chat_completion, so the
providers based on those should actually be working now.
The method `prepare_openai_completion_params` in
`llama_stack/providers/utils/inference/openai_compat.py` was improved to
actually recursively clean up input parameters, including handling of
lists, dicts, and dumping of Pydantic models to dicts. These changes
were required to get to 100% passing tests on the OpenAI API
verification against the `openai` provider.
With the above, the together.ai provider was passing as well as it is
without Llama Stack. But, since we have Llama Stack in the middle, I
took the opportunity to clean up the together.ai provider so that it now
also passes the OpenAI API spec tests we have at 100%. That means
together.ai is now passing our verification test better when using an
OpenAI client talking to Llama Stack than it is when hitting together.ai
directly, without Llama Stack in the middle.
And, another round of work for Fireworks to improve translation of
incoming OpenAI chat completion requests to Llama Stack chat completion
requests gets the fireworks provider passing at 100%. The server-side
fireworks.ai tool calling support with OpenAI chat completions and Llama
4 models isn't great yet, but by pointing the OpenAI clients at Llama
Stack's API we can clean things up and get everything working as
expected for Llama 4 models.
## Test Plan
### OpenAI API Verification Tests
I ran the OpenAI API verification tests as below and 100% of the tests
passed.
First, start a Llama Stack server that runs the `openai` provider with
the `gpt-4o` and `gpt-4o-mini` models deployed. There's not a template
setup to do this out of the box, so I added a
`tests/verifications/openai-api-verification-run.yaml` to do this.
First, ensure you have the necessary API key environment variables set:
```
export TOGETHER_API_KEY="..."
export FIREWORKS_API_KEY="..."
export OPENAI_API_KEY="..."
```
Then, run a Llama Stack server that serves up all these providers:
```
llama stack run \
--image-type venv \
tests/verifications/openai-api-verification-run.yaml
```
Finally, generate a new verification report against all these providers,
both with and without the Llama Stack server in the middle.
```
python tests/verifications/generate_report.py \
--run-tests \
--provider \
together \
fireworks \
groq \
openai \
together-llama-stack \
fireworks-llama-stack \
groq-llama-stack \
openai-llama-stack
```
You'll see that most of the configurations with Llama Stack in the
middle now pass at 100%, even though some of them do not pass at 100%
when hitting the backend provider's API directly with an OpenAI client.
### OpenAI Completion Integration Tests with vLLM:
I also ran the smaller `test_openai_completion.py` test suite (that's
not yet merged with the verification tests) on multiple of the
providers, since I had to adjust the method signature of
openai_chat_completion a bit and thus had to touch lots of these
providers to match. Here's the tests I ran there, all passing:
```
VLLM_URL="http://localhost:8000/v1" INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct" llama stack build --template remote-vllm --image-type venv --run
```
in another terminal
```
LLAMA_STACK_CONFIG=http://localhost:8321 INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "meta-llama/Llama-3.2-3B-Instruct"
```
### OpenAI Completion Integration Tests with ollama
```
INFERENCE_MODEL="llama3.2:3b-instruct-q8_0" llama stack build --template ollama --image-type venv --run
```
in another terminal
```
LLAMA_STACK_CONFIG=http://localhost:8321 INFERENCE_MODEL="llama3.2:3b-instruct-q8_0" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "llama3.2:3b-instruct-q8_0"
```
### OpenAI Completion Integration Tests with together.ai
```
INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct-Turbo" llama stack build --template together --image-type venv --run
```
in another terminal
```
LLAMA_STACK_CONFIG=http://localhost:8321 INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct-Turbo" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "meta-llama/Llama-3.2-3B-Instruct-Turbo"
```
### OpenAI Completion Integration Tests with fireworks.ai
```
INFERENCE_MODEL="meta-llama/Llama-3.1-8B-Instruct" llama stack build --template fireworks --image-type venv --run
```
in another terminal
```
LLAMA_STACK_CONFIG=http://localhost:8321 INFERENCE_MODEL="meta-llama/Llama-3.1-8B-Instruct" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "meta-llama/Llama-3.1-8B-Instruct"
---------
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
This PR adds two methods to the Inference API:
- `batch_completion`
- `batch_chat_completion`
The motivation is for evaluations targeting a local inference engine
(like meta-reference or vllm) where batch APIs provide for a substantial
amount of acceleration.
Why did I not add this to `Api.batch_inference` though? That just
resulted in a _lot_ more book-keeping given the structure of Llama
Stack. Had I done that, I would have needed to create a notion of a
"batch model" resource, setup routing based on that, etc. This does not
sound ideal.
So what's the future of the batch inference API? I am not sure. Maybe we
can keep it for true _asynchronous_ execution. So you can submit
requests, and it can return a Job instance, etc.
## Test Plan
Run meta-reference-gpu using:
```bash
export INFERENCE_MODEL=meta-llama/Llama-4-Scout-17B-16E-Instruct
export INFERENCE_CHECKPOINT_DIR=../checkpoints/Llama-4-Scout-17B-16E-Instruct-20250331210000
export MODEL_PARALLEL_SIZE=4
export MAX_BATCH_SIZE=32
export MAX_SEQ_LEN=6144
LLAMA_MODELS_DEBUG=1 llama stack run meta-reference-gpu
```
Then run the batch inference test case.
# What does this PR do?
current text for 'llama stack build' and 'llama stack run' says that if
no argument is passed to '--image-name' that the active Conda
environment will be used
in reality, the active enviroment is used whether it is from conda,
virtualenv, etc.
## Test Plan
N/A
## Documentation
N/A
Signed-off-by: Nathan Weinberg <nweinber@redhat.com>
# What does this PR do?
Today, supervised_fine_tune itself and the `TrainingConfig` class have a
bunch of required fields that a provider implementation might not need.
for example, if a provider wants to handle hyperparameters in its
configuration as well as any type of dataset retrieval, optimizer or
LoRA config, a user will still need to pass in a virtually empty
`DataConfig`, `OptimizerConfig` and `AlgorithmConfig` in some cases.
Many of these fields are intended to work specifically with llama models
and knobs intended for customizing inline.
Adding remote post_training providers will require loosening these
arguments, or forcing users to pass in empty objects to satisfy the
pydantic models.
Signed-off-by: Charlie Doern <cdoern@redhat.com>
Small docs update and an update for `start-stack.sh` with missing color
and if statment logic.
# What does this PR do?
1. Makes a small change to start-stack.sh to resolve this error:
```cmd
/home/aireilly/.local/lib/python3.13/site-packages/llama_stack/distribution/start_stack.sh: line 76: [: missing ]'
```
2. Adds a missing $GREEN colour to start-stack.sh
3. Updated `docs/source/getting_started/detailed_tutorial.md` with some
small changes and corrections.
## Test Plan
Procedures described in
`docs/source/getting_started/detailed_tutorial.md` were verified on
Linux Fedora 41.
# What does this PR do?
This stubs in some OpenAI server-side compatibility with three new
endpoints:
/v1/openai/v1/models
/v1/openai/v1/completions
/v1/openai/v1/chat/completions
This gives common inference apps using OpenAI clients the ability to
talk to Llama Stack using an endpoint like
http://localhost:8321/v1/openai/v1 .
The two "v1" instances in there isn't awesome, but the thinking is that
Llama Stack's API is v1 and then our OpenAI compatibility layer is
compatible with OpenAI V1. And, some OpenAI clients implicitly assume
the URL ends with "v1", so this gives maximum compatibility.
The openai models endpoint is implemented in the routing layer, and just
returns all the models Llama Stack knows about.
The following providers should be working with the new OpenAI
completions and chat/completions API:
* remote::anthropic (untested)
* remote::cerebras-openai-compat (untested)
* remote::fireworks (tested)
* remote::fireworks-openai-compat (untested)
* remote::gemini (untested)
* remote::groq-openai-compat (untested)
* remote::nvidia (tested)
* remote::ollama (tested)
* remote::openai (untested)
* remote::passthrough (untested)
* remote::sambanova-openai-compat (untested)
* remote::together (tested)
* remote::together-openai-compat (untested)
* remote::vllm (tested)
The goal to support this for every inference provider - proxying
directly to the provider's OpenAI endpoint for OpenAI-compatible
providers. For providers that don't have an OpenAI-compatible API, we'll
add a mixin to translate incoming OpenAI requests to Llama Stack
inference requests and translate the Llama Stack inference responses to
OpenAI responses.
This is related to #1817 but is a bit larger in scope than just chat
completions, as I have real use-cases that need the older completions
API as well.
## Test Plan
### vLLM
```
VLLM_URL="http://localhost:8000/v1" INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct" llama stack build --template remote-vllm --image-type venv --run
LLAMA_STACK_CONFIG=http://localhost:8321 INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "meta-llama/Llama-3.2-3B-Instruct"
```
### ollama
```
INFERENCE_MODEL="llama3.2:3b-instruct-q8_0" llama stack build --template ollama --image-type venv --run
LLAMA_STACK_CONFIG=http://localhost:8321 INFERENCE_MODEL="llama3.2:3b-instruct-q8_0" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "llama3.2:3b-instruct-q8_0"
```
## Documentation
Run a Llama Stack distribution that uses one of the providers mentioned
in the list above. Then, use your favorite OpenAI client to send
completion or chat completion requests with the base_url set to
http://localhost:8321/v1/openai/v1 . Replace "localhost:8321" with the
host and port of your Llama Stack server, if different.
---------
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
This PR adds unit tests for the NVIDIA Safety provider implementation.
[//]: # (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.*]
1. Ran `./scripts/unit-tests.sh
tests/unit/providers/nvidia/test_safety.py` from the root of the
project. Verified tests pass.
```
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_init_nemo_guardrails Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_init_nemo_guardrails_invalid_temperature Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_register_shield_with_valid_id Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_register_shield_without_id Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_run_shield_allowed Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_run_shield_blocked Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_run_shield_http_error Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
tests/unit/providers/nvidia/test_safety.py::TestNVIDIASafetyAdapter::test_run_shield_not_found Initializing NVIDIASafetyAdapter(http://nemo.test)...
PASSED
```
[//]: # (## Documentation)
---------
Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
# What does this PR do?
The supervised_fine_tune method in NvidiaPostTrainingAdapter had some
extra args that aren't part of the post_training protocol, and these
extra args were causing FastAPI to throw an error when attempting to
stand up an endpoint that used this provider.
(Closes#1938)
## Test Plan
Before this change, bringing up a stack with the `nvidia` template
failed. Afterwards, it passes. I'm testing this like:
```
INFERENCE_MODEL="meta/llama-3.1-8b-instruct" \
llama stack build --template nvidia --image-type venv --run
```
I also ensured the nvidia/test_supervised_fine_tuning.py tests still
pass via:
```
python -m pytest \
tests/unit/providers/nvidia/test_supervised_fine_tuning.py
```
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
This PR makes it possible to switch between agentic and non-agentic RAG
when running the respective Playground page.
When non-agentic RAG is selected, user queries are answered by directly
querying the vector DB, augmenting the prompt, and sending the extended
prompt to the model via Inference API.
## Test Plan
- Launch the Playground and go to the RAG page;
- Select the vector DB ID;
- Adjust other configuration parameters if necessary;
- Set the radio button to Agent-based RAG;
- Send a message to the chat;
- The query will be answered by an agent using the knowledge search tool
as indicated by the output;
- Click the 'Clear Chat' button to make it possible to switch modes;
- Send a message to the chat again;
- This time, the query will be answered by the model directly as can be
deduced from the reply.
# What does this PR do?
closes https://github.com/meta-llama/llama-stack/issues/1586
this issue arises when loading an mcp_endpoint from run.yaml. the issue
does not manifest for mcp servers added via a running distro server. the
existing tests only cover the case of adding to a running server.
the code for loading run.yaml strips type information from mcp_endpoint,
passing `{"uri": ...}` instead of `URL(uri=...)` along to the resource
provider registration.
## Test Plan
1. run an mcp server
2. add an mcp tool config to the dev.py, e.g.
```
diff --git a/llama_stack/templates/dev/dev.py b/llama_stack/templates/dev/dev.py
index 69924acb..e0dc7189 100644
--- a/llama_stack/templates/dev/dev.py
+++ b/llama_stack/templates/dev/dev.py
@@ -6,6 +6,8 @@
from typing import List, Tuple
+from llama_stack.apis.common.content_types import URL
+
from llama_stack.apis.models.models import ModelType
from llama_stack.distribution.datatypes import (
ModelInput,
@@ -154,6 +156,11 @@ def get_distribution_template() -> DistributionTemplate:
toolgroup_id="builtin::code_interpreter",
provider_id="code-interpreter",
),
+ ToolGroupInput(
+ toolgroup_id="mcp::filesystem",
+ provider_id="model-context-protocol",
+ mcp_endpoint=URL(uri="http://localhost:8002/sse"),
+ ),
]
embedding_model = ModelInput(
model_id="all-MiniLM-L6-v2",
```
3. run distro_codegen.py
4. llama stack build --template dev --run
before this pr, the `llama stack run` would fail w/ `AttributeError:
'dict' object has no attribute 'uri'`, after it will succeed.
# What does this PR do?
While building the "experimental-post-training" distribution, we
encountered a version conflict between torchao with inference requiring
version 0.5.0 and training currently depending on version 0.8.0.
Resolves this error:
```
× No solution found when resolving dependencies:
╰─▶ Because you require torchao==0.5.0 and torchao==0.8.0, we can conclude that your requirements are unsatisfiable.
ERROR 2025-04-10 10:41:22,597 llama_stack.distribution.build:128 uncategorized: Failed to build target test with
return code 1
```
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
This PR fixes two issues with the RAG page of the Playground UI:
1. When the user modifies a configurable setting via a widget (e.g.,
system prompt, temperature, etc.), the agent is not recreated. Thus, the
change has no effect and the user gets no indication of that.
2. After the first issue is fixed, it becomes possible to recreate the
agent mid-conversation or even mid-generation. To mitigate this, widgets
related to agent configuration are now disabled when a conversation is
in progress (i.e., when the chat is non-empty). They are automatically
enabled again when the user resets the chat history.
## Test Plan
- Launch the Playground and go to the RAG page;
- Select the vector DB ID;
- Send a message to the agent via the chat;
- The widgets in charge of the agent parameters will become disabled at
this point;
- Send a second message asking the model about the content of the first
message;
- The reply will indicate that the two messages were sent over the same
session, that is, the agent was not recreated;
- Click the 'Clear Chat' button;
- All widgets will be enabled and a new agent will be created (which can
be validated by sending another message).
Mirror to https://github.com/meta-llama/llama-models/pull/324 with some
clean up
```
with-proxy pip install -e .
export INFERENCE_MODEL=meta-llama/Llama-4-Scout-17B-16E-Instruct
export INFERENCE_CHECKPOINT_DIR=../checkpoints/Llama-4-Scout-17B-16E-Instruct
export QUANTIZATION_TYPE=int4_mixed
with-proxy llama stack build --run --template meta-reference-gpu
```
# 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)
# What does this PR do?
* Manage UI deps in pyproject
* Use a new "ui" dep group to pull the deps with "uv"
* Simplify the run command
* Bump versions in requirements.txt
Signed-off-by: Sébastien Han <seb@redhat.com>
