The non-streaming version is just a small layer on top of the streaming
version - just pluck off the final `response.completed` event and return
that as the response!
This PR also includes a couple other changes which I ended up making
while working on it on a flight:
- changes to `ollama` so it does not pull embedding models
unconditionally
- a small fix to library client to make the stream and non-stream cases
a bit more symmetric
This allows a set of rules to be defined for determining access to
resources. The rules are (loosely) based on the cedar policy format.
A rule defines a list of action either to permit or to forbid. It may
specify a principal or a resource that must match for the rule to take
effect. It may also specify a condition, either a 'when' or an 'unless',
with additional constraints as to where the rule applies.
A list of rules is held for each type to be protected and tried in order
to find a match. If a match is found, the request is permitted or
forbidden depening on the type of rule. If no match is found, the
request is denied. If no rules are specified for a given type, a rule
that allows any action as long as the resource attributes match the user
attributes is added (i.e. the previous behaviour is the default.
Some examples in yaml:
```
model:
- permit:
principal: user-1
actions: [create, read, delete]
comment: user-1 has full access to all models
- permit:
principal: user-2
actions: [read]
resource: model-1
comment: user-2 has read access to model-1 only
- permit:
actions: [read]
when:
user_in: resource.namespaces
comment: any user has read access to models with matching attributes
vector_db:
- forbid:
actions: [create, read, delete]
unless:
user_in: role::admin
comment: only user with admin role can use vector_db resources
```
---------
Signed-off-by: Gordon Sim <gsim@redhat.com>
# What does this PR do?
This adds the missing `text` parameter to the Responses API that is how
users control structured outputs. All we do with that parameter is map
it to the corresponding chat completion response_format.
## Test Plan
The new unit tests exercise the various permutations allowed for this
property, while a couple of new verification tests actually use it for
real to verify the model outputs are following the format as expected.
Unit tests:
`python -m pytest -s -v
tests/unit/providers/agents/meta_reference/test_openai_responses.py`
Verification tests:
```
llama stack run llama_stack/templates/together/run.yaml
pytest -s -vv 'tests/verifications/openai_api/test_responses.py' \
--base-url=http://localhost:8321/v1/openai/v1 \
--model meta-llama/Llama-4-Scout-17B-16E-Instruct
```
Note that the verification tests can only be run with a real Llama Stack
server (as opposed to using the library client via
`--provider=stack:together`) because the Llama Stack python client is
not yet updated to accept this text field.
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
TSIA
Added Files provider to the fireworks template. Might want to add to all
templates as a follow-up.
## Test Plan
llama-stack pytest tests/unit/files/test_files.py
llama-stack llama stack build --template fireworks --image-type conda
--run
LLAMA_STACK_CONFIG=http://localhost:8321 pytest -s -v
tests/integration/files/
# What does this PR do?
The chat completion ids generated by Ollama are not unique enough to use
with stored chat completions as they rely on only 3 numbers of
randomness to give unique values - ie `chatcmpl-373`. This causes
frequent collisions in id values of chat completions in Ollama, which
creates issues in our SQL storage of chat completions by id where it
expects ids to actually be unique.
So, this adjusts Ollama responses to use uuids as unique ids. This does
mean we're replacing the ids generated natively by Ollama. If we don't
wish to do this, we'll either need to relax the unique constraint on our
chat completions id field in the inference storage or convince Ollama
upstream to use something closer to uuid values here.
Closes#2315
## Test Plan
I tested by running the openai completion / chat completion integration
tests in a loop. Without this change, I regularly get unique id
collisions. With this change, I do not. We sometimes see flakes from
these unique id collisions in our CI tests, and this will resolve those.
```
INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct" \
llama stack run llama_stack/templates/ollama/run.yaml
while true; do; \
INFERENCE_MODEL="meta-llama/Llama-3.2-3B-Instruct" \
pytest -s -v \
tests/integration/inference/test_openai_completion.py \
--stack-config=http://localhost:8321 \
--text-model="meta-llama/Llama-3.2-3B-Instruct"; \
done
```
Signed-off-by: Ben Browning <bbrownin@redhat.com>
I think the implementation needs more simplification. Spent way too much
time trying to get the tests pass with models not co-operating :(
Finally had to switch claude-sonnet to get things to pass reliably.
### Test Plan
```
export TAVILY_SEARCH_API_KEY=...
export OPENAI_API_KEY=...
uv run pytest -p no:warnings \
-s -v tests/verifications/openai_api/test_responses.py \
--provider=stack:starter \
--model openai/gpt-4o
```
fixes provider to use stream var correctly
Before
```
curl --request POST \
--url http://localhost:8321/v1/openai/v1/chat/completions \
--header 'content-type: application/json' \
--data '{
"model": "meta-llama/Llama-4-Scout-17B-16E-Instruct",
"messages": [
{
"role": "user",
"content": "Who are you?"
}
]
}'
{"detail":"Internal server error: An unexpected error occurred."}
```
After
```
llama-stack % curl --request POST \
--url http://localhost:8321/v1/openai/v1/chat/completions \
--header 'content-type: application/json' \
--data '{
"model": "accounts/fireworks/models/llama4-scout-instruct-basic",
"messages": [
{
"role": "user",
"content": "Who are you?"
}
]
}'
{"id":"chatcmpl-97978538-271d-4c73-8d4d-c509bfb6c87e","choices":[{"message":{"role":"assistant","content":"I'm an AI assistant designed by Meta. I'm here to answer your questions, share interesting ideas and maybe even surprise you with a fresh perspective. What's on your mind?","name":null,"tool_calls":null},"finish_reason":"stop","index":0,"logprobs":null}],"object":"chat.completion","created":1748896403,"model":"accounts/fireworks/models/llama4-scout-instruct-basic"}%
```
# What does this PR do?
Adds a new endpoint that is compatible with OpenAI for embeddings api.
`/openai/v1/embeddings`
Added providers for OpenAI, LiteLLM and SentenceTransformer.
## Test Plan
```
LLAMA_STACK_CONFIG=http://localhost:8321 pytest -sv tests/integration/inference/test_openai_embeddings.py --embedding-model all-MiniLM-L6-v2,text-embedding-3-small,gemini/text-embedding-004
```
# What does this PR do?
This adds a check to ensure we don't attempt to concatenate `None + str`
or `str + None` when building up our arguments for streaming tool calls
in the Responses API.
## Test Plan
All existing tests pass with this change.
Unit tests:
```
python -m pytest -s -v \
tests/unit/providers/agents/meta_reference/test_openai_responses.py
```
Integration tests:
```
llama stack run llama_stack/templates/together/run.yaml
LLAMA_STACK_CONFIG=http://localhost:8321 \
python -m pytest -s -v \
tests/integration/agents/test_openai_responses.py \
--text-model meta-llama/Llama-4-Scout-17B-16E-Instruct
```
Verification tests:
```
llama stack run llama_stack/templates/together/run.yaml
pytest -s -v 'tests/verifications/openai_api/test_responses.py' \
--base-url=http://localhost:8321/v1/openai/v1 \
--model meta-llama/Llama-4-Scout-17B-16E-Instruct
```
Additionally, the manual example using Codex CLI from #2325 now succeeds
instead of throwing a 500 error.
Closes#2325
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
* Added support postgresql inference store
* Added 'oracle' template that demos how to config postgresql stores
(except for telemetry, which is not supported currently)
## Test Plan
llama stack build --template oracle --image-type conda --run
LLAMA_STACK_CONFIG=http://localhost:8321 pytest -s -v tests/integration/
--text-model accounts/fireworks/models/llama-v3p3-70b-instruct -k
'inference_store'
# What does this PR do?
Updates sambanova inference to use strict as false in json_schema
structured output
## Test Plan
pytest -s -v tests/integration/inference/test_text_inference.py
--stack-config=sambanova
--text-model=sambanova/Meta-Llama-3.3-70B-Instruct
We must store the full (re-hydrated) input not just the original input
in the Response object. Of course, this is not very space efficient and
we should likely find a better storage scheme so that we can only store
unique entries in the database and then re-hydrate them efficiently
later. But that can be done safely later.
Closes https://github.com/meta-llama/llama-stack/issues/2299
## Test Plan
Unit test
# What does this PR do?
Previously prompt guard was hard coded to require cuda which prevented
it from being used on an instance without a cuda support.
This PR allows prompt guard to be configured to use either cpu or cuda.
[//]: # (If resolving an issue, uncomment and update the line below)
Closes [#2133](https://github.com/meta-llama/llama-stack/issues/2133)
## Test Plan (Edited after incorporating suggestion)
1) started stack configured with prompt guard as follows on a system
without a GPU
and validated prompt guard could be used through the APIs
2) validated on a system with a gpu (but without llama stack) that the
python selecting between cpu and cuda support returned the right value
when a cuda device was available.
3) ran the unit tests as per -
https://github.com/meta-llama/llama-stack/blob/main/tests/unit/README.md
[//]: # (## Documentation)
---------
Signed-off-by: Michael Dawson <mdawson@devrus.com>
# What does this PR do?
Fix a bug in openai_compat where choices are not indexed correctly.
## Test Plan
Added a new test.
Rerun the failed inference_store tests:
llama stack run fireworks --image-type conda
pytest -s -v tests/integration/ --stack-config http://localhost:8321 -k
'test_inference_store' --text-model meta-llama/Llama-3.3-70B-Instruct
--count 10
This adds initial streaming support to the Responses API.
This PR makes sure that the _first_ inference call made to chat
completions streams out.
There's more to be done:
- tool call output tokens need to stream out when possible
- we need to loop through multiple rounds of inference and they all need
to stream out.
## Test Plan
Added a test. Executed as:
```
FIREWORKS_API_KEY=... \
pytest -s -v 'tests/verifications/openai_api/test_responses.py' \
--provider=stack:fireworks --model meta-llama/Llama-4-Scout-17B-16E-Instruct
```
Then, started a llama stack fireworks distro and tested against it like
this:
```
OPENAI_API_KEY=blah \
pytest -s -v 'tests/verifications/openai_api/test_responses.py' \
--base-url http://localhost:8321/v1/openai/v1 \
--model meta-llama/Llama-4-Scout-17B-16E-Instruct
```
# What does this PR do?
Handles the case where the vllm config `tls_verify` is set to `false` or
`true`.
Closes: https://github.com/meta-llama/llama-stack/issues/2283
Signed-off-by: Sébastien Han <seb@redhat.com>
# What does this PR do?
The `tls_verify` can now receive a path to a certificate file if the
endpoint requires it.
Signed-off-by: Sébastien Han <seb@redhat.com>
When registering a MCP endpoint, we cannot list tools (like we used to)
since the MCP endpoint may be behind an auth wall. Registration can
happen much sooner (via run.yaml).
Instead, we do listing only when the _user_ actually calls listing.
Furthermore, we cache the list in-memory in the server. Currently, the
cache is not invalidated -- we may want to periodically re-list for MCP
servers. Note that they must call `list_tools` before calling
`invoke_tool` -- we use this critically.
This will enable us to list MCP servers in run.yaml
## Test Plan
Existing tests, updated tests accordingly.
# What does this PR do?
This is not part of the official OpenAI API, but we'll use this for the
logs UI.
In order to support more filtering options, I'm adopting the newly
introduced sql store in in place of the kv store.
## Test Plan
Added integration/unit tests.
The most interesting MCP servers are those with an authorization wall in
front of them. This PR uses the existing `provider_data` mechanism of
passing provider API keys for passing MCP access tokens (in fact,
arbitrary headers in the style of the OpenAI Responses API) from the
client through to the MCP server.
```
class MCPProviderDataValidator(BaseModel):
# mcp_endpoint => list of headers to send
mcp_headers: dict[str, list[str]] | None = None
```
Note how we must stuff the headers for all MCP endpoints into a single
"MCPProviderDataValidator". Unlike existing providers (e.g., Together
and Fireworks for inference) where we could name the provider api keys
clearly (`together_api_key`, `fireworks_api_key`), we cannot name these
keys for MCP. We have a single generic MCP provider which can serve
multiple "toolgroups". So we use a dict to combine all the headers for
all MCP endpoints you may want to use in an agentic call.
## Test Plan
See the added integration test for usage.
# What does this PR do?
Since https://github.com/meta-llama/llama-stack/pull/2193 switched to
openai sdk, we need to strip 'openai/' from the model_id
## Test Plan
start server with openai provider and send a chat completion call
# What does this PR do?
* Provide sqlite implementation of the APIs introduced in
https://github.com/meta-llama/llama-stack/pull/2145.
* Introduced a SqlStore API: llama_stack/providers/utils/sqlstore/api.py
and the first Sqlite implementation
* Pagination support will be added in a future PR.
## Test Plan
Unit test on sql store:
<img width="1005" alt="image"
src="https://github.com/user-attachments/assets/9b8b7ec8-632b-4667-8127-5583426b2e29"
/>
Integration test:
```
INFERENCE_MODEL="llama3.2:3b-instruct-fp16" llama stack build --template ollama --image-type conda --run
```
```
LLAMA_STACK_CONFIG=http://localhost:5001 INFERENCE_MODEL="llama3.2:3b-instruct-fp16" python -m pytest -v tests/integration/inference/test_openai_completion.py --text-model "llama3.2:3b-instruct-fp16" -k 'inference_store and openai'
```
# What does this PR do?
Includes SambaNova safety adaptor to use the sambanova cloud served
Meta-Llama-Guard-3-8B
minor updates in sambanova docs
## Test Plan
pytest -s -v tests/integration/safety/test_safety.py
--stack-config=sambanova --safety-shield=sambanova/Meta-Llama-Guard-3-8B
# What does this PR do?
This PR introduces support for keyword based FTS5 search with BM25
relevance scoring. It makes changes to the existing EmbeddingIndex base
class in order to support a search_mode and query_str parameter, that
can be used for keyword based search implementations.
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
run
```
pytest llama_stack/providers/tests/vector_io/test_sqlite_vec.py -v -s --tb=short --disable-warnings --asyncio-mode=auto
```
Output:
```
pytest llama_stack/providers/tests/vector_io/test_sqlite_vec.py -v -s --tb=short --disable-warnings --asyncio-mode=auto
/Users/vnarsing/miniconda3/envs/stack-client/lib/python3.10/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 3.10.16, pytest-8.3.4, pluggy-1.5.0 -- /Users/vnarsing/miniconda3/envs/stack-client/bin/python
cachedir: .pytest_cache
metadata: {'Python': '3.10.16', 'Platform': 'macOS-14.7.4-arm64-arm-64bit', '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'}}
rootdir: /Users/vnarsing/go/src/github/meta-llama/llama-stack
configfile: pyproject.toml
plugins: html-4.1.1, metadata-3.1.1, asyncio-0.25.3, anyio-4.8.0
asyncio: mode=auto, asyncio_default_fixture_loop_scope=None
collected 7 items
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_add_chunks PASSED
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_query_chunks_vector PASSED
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_query_chunks_fts PASSED
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_chunk_id_conflict PASSED
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_register_vector_db PASSED
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_unregister_vector_db PASSED
llama_stack/providers/tests/vector_io/test_sqlite_vec.py::test_generate_chunk_id PASSED
```
For reference, with the implementation, the fts table looks like below:
```
Chunk ID: 9fbc39ce-c729-64a2-260f-c5ec9bb2a33e, Content: Sentence 0 from document 0
Chunk ID: 94062914-3e23-44cf-1e50-9e25821ba882, Content: Sentence 1 from document 0
Chunk ID: e6cfd559-4641-33ba-6ce1-7038226495eb, Content: Sentence 2 from document 0
Chunk ID: 1383af9b-f1f0-f417-4de5-65fe9456cc20, Content: Sentence 3 from document 0
Chunk ID: 2db19b1a-de14-353b-f4e1-085e8463361c, Content: Sentence 4 from document 0
Chunk ID: 9faf986a-f028-7714-068a-1c795e8f2598, Content: Sentence 5 from document 0
Chunk ID: ef593ead-5a4a-392f-7ad8-471a50f033e8, Content: Sentence 6 from document 0
Chunk ID: e161950f-021f-7300-4d05-3166738b94cf, Content: Sentence 7 from document 0
Chunk ID: 90610fc4-67c1-e740-f043-709c5978867a, Content: Sentence 8 from document 0
Chunk ID: 97712879-6fff-98ad-0558-e9f42e6b81d3, Content: Sentence 9 from document 0
Chunk ID: aea70411-51df-61ba-d2f0-cb2b5972c210, Content: Sentence 0 from document 1
Chunk ID: b678a463-7b84-92b8-abb2-27e9a1977e3c, Content: Sentence 1 from document 1
Chunk ID: 27bd63da-909c-1606-a109-75bdb9479882, Content: Sentence 2 from document 1
Chunk ID: a2ad49ad-f9be-5372-e0c7-7b0221d0b53e, Content: Sentence 3 from document 1
Chunk ID: cac53bcd-1965-082a-c0f4-ceee7323fc70, Content: Sentence 4 from document 1
```
Query results:
Result 1: Sentence 5 from document 0
Result 2: Sentence 5 from document 1
Result 3: Sentence 5 from document 2
[//]: # (## Documentation)
---------
Signed-off-by: Varsha Prasad Narsing <varshaprasad96@gmail.com>
# What does this PR do?
Add support for "instructions" to the responses API. Instructions
provide a way to swap out system (or developer) messages in new
responses.
## Test Plan
unit tests added
Signed-off-by: Derek Higgins <derekh@redhat.com>
# What does this PR do?
When launching a fine-tuning job, an upcoming version of NeMo Customizer
will expect the `config` name to be formatted as
`namespace/name@version`. Here, `config` is a reference to a model +
additional metadata. There could be multiple `config`s that reference
the same base model.
This PR updates NVIDIA's `supervised_fine_tune` to simply pass the
`model` param as-is to NeMo Customizer. Currently, it expects a
specific, allowlisted llama model (i.e. `meta/Llama3.1-8B-Instruct`) and
converts it to the provider format (`meta/llama-3.1-8b-instruct`).
[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])
## Test Plan
From a notebook, I built an image with my changes:
```
!llama stack build --template nvidia --image-type venv
from llama_stack.distribution.library_client import LlamaStackAsLibraryClient
client = LlamaStackAsLibraryClient("nvidia")
client.initialize()
```
And could successfully launch a job:
```
response = client.post_training.supervised_fine_tune(
job_uuid="",
model="meta/llama-3.2-1b-instruct@v1.0.0+A100", # Model passed as-is to Customimzer
...
)
job_id = response.job_uuid
print(f"Created job with ID: {job_id}")
Output:
Created job with ID: cust-Jm4oGmbwcvoufaLU4XkrRU
```
[//]: # (## Documentation)
---------
Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
# What does this PR do?
If a user has previously serialized data into their vector store without
the `metadata_token_count` in the chunk, the `query` method will fail in
a server error. This fixes that edge case by returning 0 when the key is
not detected. This solution is suboptimal but I think it's better to
understate the token size rather than recalculate it and add unnecessary
complexity to the retrieval code.
[//]: # (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)
Signed-off-by: Francisco Javier Arceo <farceo@redhat.com>
# What does this PR do?
This extracts the W3C trace context headers (traceparent and tracestate)
from incoming requests, stuffs them as attributes on the spans we
create, and uses them within the tracing provider implementation to
actually wrap our spans in the proper context.
What this means in practice is that when a client (such as an OpenAI
client) is instrumented to create these traces, we'll continue that
distributed trace within Llama Stack as opposed to creating our own root
span that breaks the distributed trace between client and server.
It's slightly awkward to do this in Llama Stack because our Tracing API
knows nothing about opentelemetry, W3C trace headers, etc - that's only
knowledge the specific provider implementation has. So, that's why the
trace headers get extracted by in the server code but not actually used
until the provider implementation to form the proper context.
This also centralizes how we were adding the `__root__` and
`__root_span__` attributes, as those two were being added in different
parts of the code instead of from a single place.
Closes#2097
## Test Plan
This was tested manually using the helpful scripts from #2097. I
verified that Llama Stack properly joined the client's span when the
client was instrumented for distributed tracing, and that Llama Stack
properly started its own root span when the incoming request was not
part of an existing trace.
Here's an example of the joined spans:
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
This PR introduces APIs to retrieve past chat completion requests, which
will be used in the LS UI.
Our current `Telemetry` is ill-suited for this purpose as it's untyped
so we'd need to filter by obscure attribute names, making it brittle.
Since these APIs are 'provided by stack' and don't need to be
implemented by inference providers, we introduce a new InferenceProvider
class, containing the existing inference protocol, which is implemented
by inference providers.
The APIs are OpenAI-compliant, with an additional `input_messages`
field.
## Test Plan
This PR just adds the API and marks them provided_by_stack. S
tart stack server -> doesn't crash
# What does this PR do?
adds an inline HF SFTTrainer provider. Alongside touchtune -- this is a
super popular option for running training jobs. The config allows a user
to specify some key fields such as a model, chat_template, device, etc
the provider comes with one recipe `finetune_single_device` which works
both with and without LoRA.
any model that is a valid HF identifier can be given and the model will
be pulled.
this has been tested so far with CPU and MPS device types, but should be
compatible with CUDA out of the box
The provider processes the given dataset into the proper format,
establishes the various steps per epoch, steps per save, steps per eval,
sets a sane SFTConfig, and runs n_epochs of training
if checkpoint_dir is none, no model is saved. If there is a checkpoint
dir, a model is saved every `save_steps` and at the end of training.
## Test Plan
re-enabled post_training integration test suite with a singular test
that loads the simpleqa dataset:
https://huggingface.co/datasets/llamastack/simpleqa and a tiny granite
model: https://huggingface.co/ibm-granite/granite-3.3-2b-instruct. The
test now uses the llama stack client and the proper post_training API
runs one step with a batch_size of 1. This test runs on CPU on the
Ubuntu runner so it needs to be a small batch and a single step.
[//]: # (## Documentation)
---------
Signed-off-by: Charlie Doern <cdoern@redhat.com>
# What does this PR do?
fixes#2121
this implementation splits reponsibility between litellm and openai
libraries -
| Inference Method | Implementation Source |
|----------------------------|--------------------------|
| completion | LiteLLMOpenAIMixin |
| chat_completion | LiteLLMOpenAIMixin |
| embedding | LiteLLMOpenAIMixin |
| batch_completion | LiteLLMOpenAIMixin |
| batch_chat_completion | LiteLLMOpenAIMixin |
| openai_completion | AsyncOpenAI |
| openai_chat_completion | AsyncOpenAI |
## Test Plan
smoke test with -
```
$ OPENAI_API_KEY=$LLAMA_API_KEY OPENAI_BASE_URL=https://api.llama.com/compat/v1 llama stack build --image-type conda --image-name openai --providers inference=remote::openai --run
$ llama-stack-client models register Llama-4-Scout-17B-16E-Instruct-FP8
$ curl "http://localhost:8321/v1/openai/v1/chat/completions" -H "Content-Type: application/json" \ -d '{
"model": "Llama-4-Scout-17B-16E-Instruct-FP8",
"messages": [
{"role": "user", "content": "Hello Llama! Can you give me a quick intro?"}
]
}'
{"id":"AmPwrrkc5JgVjejPdIPrpT2","choices":[{"finish_reason":"stop","index":0,"logprobs":{"content":null,"refusal":null},"message":{"content":"Hello! I'm Llama, a Meta-designed model that adapts to your conversational style. Whether you need quick answers, deep dives into ideas, or just want to vent, joke, or brainstorm—I'm here for it. What’s on your mind?","refusal":"","role":"assistant","annotations":null,"audio":null,"function_call":null,"tool_calls":null,"id":"AmPwrrkc5JgVjejPdIPrpT2"}}],"created":1747410061,"model":"Llama-4-Scout-17B-16E-Instruct-FP8","object":"chat.completions","service_tier":null,"system_fingerprint":null,"usage":{"completion_tokens":54,"prompt_tokens":22,"total_tokens":76,"completion_tokens_details":null,"prompt_tokens_details":null}}
```
and run full test suite.
# What does this PR do?
This fixes an issue in how we used the tool_call_buf from streaming tool
calls in the remote-vllm provider where it would end up concatenating
parameters from multiple different tool call results instead of
aggregating the results from each tool call separately.
It also fixes an issue found while digging into that where we were
accidentally mixing the json string form of tool call parameters with
the string representation of the python form, which mean we'd end up
with single quotes in what should be double-quoted json strings.
Closes#1120
## Test Plan
The following tests are now passing 100% for the remote-vllm provider,
where some of the test_text_inference were failing before this change:
```
VLLM_URL="http://localhost:8000/v1" INFERENCE_MODEL="RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic" LLAMA_STACK_CONFIG=remote-vllm python -m pytest -v tests/integration/inference/test_text_inference.py --text-model "RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic"
VLLM_URL="http://localhost:8000/v1" INFERENCE_MODEL="RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic" LLAMA_STACK_CONFIG=remote-vllm python -m pytest -v tests/integration/inference/test_vision_inference.py --vision-model "RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic"
```
All but one of the agent tests are passing (including the multi-tool
one). See the PR at https://github.com/vllm-project/vllm/pull/17917 and
a gist at
https://gist.github.com/bbrowning/4734240ce96b4264340caa9584e47c9e for
changes needed there, which will have to get made upstream in vLLM.
Agent tests:
```
VLLM_URL="http://localhost:8000/v1" INFERENCE_MODEL="RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic" LLAMA_STACK_CONFIG=remote-vllm python -m pytest -v tests/integration/agents/test_agents.py --text-model "RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic"
````
---------
Signed-off-by: Ben Browning <bbrownin@redhat.com>
# What does this PR do?
his PR allows users to customize the template used for chunks when
inserted into the context. Additionally, this enables metadata injection
into the context of an LLM for RAG. This makes a naive and crude
assumption that each chunk should include the metadata, this is
obviously redundant when multiple chunks are returned from the same
document. In order to remove any sort of duplication of chunks, we'd
have to make much more significant changes so this is a reasonable first
step that unblocks users requesting this enhancement in
https://github.com/meta-llama/llama-stack/issues/1767.
In the future, this can be extended to support citations.
List of Changes:
- `llama_stack/apis/tools/rag_tool.py`
- Added `chunk_template` field in `RAGQueryConfig`.
- Added `field_validator` to validate the `chunk_template` field in
`RAGQueryConfig`.
- Ensured the `chunk_template` field includes placeholders `{index}` and
`{chunk.content}`.
- Updated the `query` method to use the `chunk_template` for formatting
chunk text content.
- `llama_stack/providers/inline/tool_runtime/rag/memory.py`
- Modified the `insert` method to pass `doc.metadata` for chunk
creation.
- Enhanced the `query` method to format results using `chunk_template`
and exclude unnecessary metadata fields like `token_count`.
- `llama_stack/providers/utils/memory/vector_store.py`
- Updated `make_overlapped_chunks` to include metadata serialization and
token count for both content and metadata.
- Added error handling for metadata serialization issues.
- `pyproject.toml`
- Added `pydantic.field_validator` as a recognized `classmethod`
decorator in the linting configuration.
- `tests/integration/tool_runtime/test_rag_tool.py`
- Refactored test assertions to separate `assert_valid_chunk_response`
and `assert_valid_text_response`.
- Added integration tests to validate `chunk_template` functionality
with and without metadata inclusion.
- Included a test case to ensure `chunk_template` validation errors are
raised appropriately.
- `tests/unit/rag/test_vector_store.py`
- Added unit tests for `make_overlapped_chunks`, verifying chunk
creation with overlapping tokens and metadata integrity.
- Added tests to handle metadata serialization errors, ensuring proper
exception handling.
- `docs/_static/llama-stack-spec.html`
- Added a new `chunk_template` field of type `string` with a default
template for formatting retrieved chunks in RAGQueryConfig.
- Updated the `required` fields to include `chunk_template`.
- `docs/_static/llama-stack-spec.yaml`
- Introduced `chunk_template` field with a default value for
RAGQueryConfig.
- Updated the required configuration list to include `chunk_template`.
- `docs/source/building_applications/rag.md`
- Documented the `chunk_template` configuration, explaining how to
customize metadata formatting in RAG queries.
- Added examples demonstrating the usage of the `chunk_template` field
in RAG tool queries.
- Highlighted default values for `RAG` agent configurations.
# Resolves https://github.com/meta-llama/llama-stack/issues/1767
## Test Plan
Updated both `test_vector_store.py` and `test_rag_tool.py` and tested
end-to-end with a script.
I also tested the quickstart to enable this and specified this metadata:
```python
document = RAGDocument(
document_id="document_1",
content=source,
mime_type="text/html",
metadata={"author": "Paul Graham", "title": "How to do great work"},
)
```
Which produced the output below:
This highlights the usefulness of the additional metadata. Notice how
the metadata is redundant for different chunks of the same document. I
think we can update that in a subsequent PR.
# Documentation
I've added a brief comment about this in the documentation to outline
this to users and updated the API documentation.
---------
Signed-off-by: Francisco Javier Arceo <farceo@redhat.com>
# What does this PR do?
In the Responses API, we convert incoming response requests to chat
completion requests. When streaming the resulting chunks of those chat
completion requests, inference providers that use OpenAI clients will
often return a `type=None` value in the tool call parts of the response.
This causes issues when we try to dump and load that response into our
pydantic model, because type cannot be None in the Responses API model
we're loading these into.
So, strip the "type" field, if present, off those chat completion tool
call results before dumping and loading them as our typed pydantic
models, which will apply our default value for that type field.
## Test Plan
This was found via manual testing of the Responses API with codex, where
I was getting errors in some tool call situations. I added a unit test
to simulate this scenario and verify the fix, as well as manual codex
testing to verify the fix.
Signed-off-by: Ben Browning <bbrownin@redhat.com>
note: the openai provider exposes the litellm specific model names to
the user. this change is compatible with that. the litellm names should
be deprecated.
# What does this PR do?
Closes#2113.
Closes#1783.
Fixes a bug in handling the end of tool execution request stream where
no `finish_reason` is provided by the model.
## Test Plan
1. Ran existing unit tests
2. Added a dedicated test verifying correct behavior in this edge case
3. Ran the code snapshot from #2113
[//]: # (## Documentation)
# What does this PR do?
Closes#2111.
Fixes an error causing Llama Stack to just return `<tool_call>` and
complete the turn without actually executing the tool. See the issue
description for more detail.
## Test Plan
1) Ran existing unit tests
2) Added a dedicated test verifying correct behavior in this edge case
3) Ran the code snapshot from #2111
