73 commits

Author	SHA1	Message	Date
Matthew Farrellee	477bcd4d09	feat: allow dynamic model registration for nvidia inference provider (#2726) ... # What does this PR do? let's users register models available at https://integrate.api.nvidia.com/v1/models that isn't already in llama_stack/providers/remote/inference/nvidia/models.py ## Test Plan 1. run the nvidia distro 2. register a model from https://integrate.api.nvidia.com/v1/models that isn't already know, as of this writing nvidia/llama-3.1-nemotron-ultra-253b-v1 is a good example 3. perform inference w/ the model	2025-07-17 12:11:30 -07:00
Sergey Yedrikov	30be1fd8b7	fix: SQLiteVecIndex.create(..., bank_id="test_bank.123") - bank_id with a dot - leads to sqlite3.OperationalError (#2770) (#2771) ... # What does this PR do? Resolves https://github.com/meta-llama/llama-stack/issues/2770. It replaces characters in SQLite table names that are not alphanumeric or underscores with underscores and quotes the table names with square brackets in SQL statements. Closes #[2770] ## Test Plan I added a ".123" suffix to the bank_id on the following line ``` index = await SQLiteVecIndex.create(dimension=embedding_dimension, db_path=db_path, bank_id="test_bank.123") ``` in tests/unit/providers/vector_io/test_sqlite_vec.py, which, without the fix in place, demonstrates the issue.	2025-07-16 08:25:44 -07:00
Matthew Farrellee	a3e249807b	chore: remove vision model URL workarounds and simplify client creation (#2775) ... The vision models are now available at the standard URL, so the workaround code has been removed. This also simplifies the codebase by eliminating the need for per-model client caching. - Remove special URL handling for meta/llama-3.2-11b/90b-vision-instruct models - Convert _get_client method to _client property for cleaner API - Remove unnecessary lru_cache decorator and functools import - Simplify client creation logic to use single base URL for all models	2025-07-16 07:10:04 -07:00
Varsha	4ae5656c2f	feat: Implement keyword search in milvus (#2231) ... # What does this PR do? This PR adds the keyword search implementation for Milvus. Along with the implementation for remote Milvus, the tests require us to start a Milvus containers locally. In order to verify the implementation, run: ``` pytest tests/unit/providers/vector_io/remote/test_milvus.py -v -s --tb=short --disable-warnings --asyncio-mode=auto ``` You can also test the changes using the below script: ``` #!/usr/bin/env python3 import asyncio import os import uuid from typing import List from llama_stack_client import ( Agent, AgentEventLogger, LlamaStackClient, RAGDocument ) class MilvusRAGDemo: def __init__(self, base_url: str = "http://localhost:8321/"): self.client = LlamaStackClient(base_url=base_url) self.vector_db_id = f"milvus_rag_demo_{uuid.uuid4().hex[:8]}" self.model_id = None self.embedding_model_id = None self.embedding_dimension = None def setup_models(self): """Get available models and select appropriate ones for LLM and embeddings.""" models = self.client.models.list() # Select embedding model embedding_models = [m for m in models if m.model_type == "embedding"] if not embedding_models: raise ValueError("No embedding models found") self.embedding_model_id = embedding_models[0].identifier self.embedding_dimension = embedding_models[0].metadata["embedding_dimension"] def register_vector_db(self): print(f"Registering Milvus vector database: {self.vector_db_id}") response = self.client.vector_dbs.register( vector_db_id=self.vector_db_id, embedding_model=self.embedding_model_id, embedding_dimension=self.embedding_dimension, provider_id="milvus-remote", # Use remote Milvus ) print(f"Vector database registered successfully") return response def insert_documents(self): """Insert sample documents into the vector database.""" print("\nInserting sample documents...") # Sample documents about different topics documents = [ RAGDocument( document_id="ai_ml_basics", content=""" Artificial Intelligence (AI) and Machine Learning (ML) are transforming the world. AI refers to the simulation of human intelligence in machines, while ML is a subset of AI that enables computers to learn and improve from experience without being explicitly programmed. Deep learning, a subset of ML, uses neural networks with multiple layers to process complex patterns in data. Key concepts in AI/ML include: - Supervised Learning: Training with labeled data - Unsupervised Learning: Finding patterns in unlabeled data - Reinforcement Learning: Learning through trial and error - Neural Networks: Computing systems inspired by biological brains """, mime_type="text/plain", metadata={"topic": "technology", "category": "ai_ml"}, ), ] # Insert documents with chunking self.client.tool_runtime.rag_tool.insert( documents=documents, vector_db_id=self.vector_db_id, chunk_size_in_tokens=200, # Smaller chunks for better granularity ) print(f"Inserted {len(documents)} documents with chunking") def test_keyword_search(self): """Test keyword-based search using BM25.""" queries = [ "neural networks", "Python frameworks", "data cleaning", ] for query in queries: response = self.client.vector_io.query( vector_db_id=self.vector_db_id, query=query, params={ "mode": "keyword", # Keyword search "max_chunks": 3, "score_threshold": 0.0, } ) for i, (chunk, score) in enumerate(zip(response.chunks, response.scores)): print(f" {i+1}. Score: {score:.4f}") print(f" Content: {chunk.content[:100]}...") print(f" Metadata: {chunk.metadata}") def run_demo(self): try: self.setup_models() self.register_vector_db() self.insert_documents() self.test_keyword_search() except Exception as e: print(f"Error during demo: {e}") raise def main(): """Main function to run the demo.""" # Check if Llama Stack server is running demo = MilvusRAGDemo() try: demo.run_demo() except Exception as e: print(f"Demo failed: {e}") if __name__ == "__main__": main() ``` [//]: # (## Documentation) --------- Signed-off-by: Varsha Prasad Narsing <varshaprasad96@gmail.com>	2025-07-14 19:39:55 -04:00
Matthew Farrellee	f731f369a2	feat: add infrastructure to allow inference model discovery (#2710) ... # What does this PR do? inference providers each have a static list of supported / known models. some also have access to a dynamic list of currently available models. this change gives prodivers using the ModelRegistryHelper the ability to combine their static and dynamic lists. for instance, OpenAIInferenceAdapter can implement ``` def query_available_models(self) -> list[str]: return [entry.model for entry in self.openai_client.models.list()] ``` to augment its static list w/ a current list from openai. ## Test Plan scripts/unit-test.sh	2025-07-14 11:38:53 -07:00
Derek Higgins	a7ed86181c	fix(faiss): Delete file contents from kvstore (#2686) ... Remove both the metadata and content from the kvstore when a file is being removed from the vector store. Closes: #2685 Also add faiss provider to openai_vector_stores test suite --------- Signed-off-by: Derek Higgins <derekh@redhat.com> Co-authored-by: raghotham <rsm@meta.com>	2025-07-14 13:58:23 -04:00
Matthew Farrellee	68e7978c88	chore: block network access from unit tests (#2732) ... # What does this PR do? this blocks network access for all `tests/unit/` tests. `tests/integration/` are untouched. it also introduces an `allow_network` marker to explicitly allow network access. ## Test Plan `./scripts/unit-tests.sh`	2025-07-12 16:53:54 -07:00
Ben Browning	51d9fd4808	fix: Don't cache clients for passthrough auth providers (#2728) ... # What does this PR do? Some of our inference providers support passthrough authentication via `x-llamastack-provider-data` header values. This fixes the providers that support passthrough auth to not cache their clients to the backend providers (mostly OpenAI client instances) so that the client connecting to Llama Stack has to provide those auth values on each and every request. ## Test Plan I added some unit tests to ensure we're not caching clients across requests for all the fixed providers in this PR. ``` uv run pytest -sv tests/unit/providers/inference/test_inference_client_caching.py ``` I also ran some of our OpenAI compatible API integration tests for each of the changed providers, just to ensure they still work. Note that these providers don't actually pass all these tests (for unrelated reasons due to quirks of the Groq and Together SaaS services), but enough of the tests passed to confirm the clients are still working as intended. ### Together ``` ENABLE_TOGETHER="together" \ uv run llama stack run llama_stack/templates/starter/run.yaml LLAMA_STACK_CONFIG=http://localhost:8321 \ uv run pytest -sv \ tests/integration/inference/test_openai_completion.py \ --text-model "together/meta-llama/Llama-3.1-8B-Instruct" ``` ### OpenAI ``` ENABLE_OPENAI="openai" \ uv run llama stack run llama_stack/templates/starter/run.yaml LLAMA_STACK_CONFIG=http://localhost:8321 \ uv run pytest -sv \ tests/integration/inference/test_openai_completion.py \ --text-model "openai/gpt-4o-mini" ``` ### Groq ``` ENABLE_GROQ="groq" \ uv run llama stack run llama_stack/templates/starter/run.yaml LLAMA_STACK_CONFIG=http://localhost:8321 \ uv run pytest -sv \ tests/integration/inference/test_openai_completion.py \ --text-model "groq/meta-llama/Llama-3.1-8B-Instruct" ``` --------- Signed-off-by: Ben Browning <bbrownin@redhat.com>	2025-07-11 13:38:27 -07:00
Matthew Farrellee	30b2e6a495	chore: default to pytest asyncio-mode=auto (#2730) ... # What does this PR do? previously, developers who ran `./scripts/unit-tests.sh` would get `asyncio-mode=auto`, which meant `@pytest.mark.asyncio` and `@pytest_asyncio.fixture` were redundent. developers who ran `pytest` directly would get pytest's default (strict mode), would run into errors leading them to add `@pytest.mark.asyncio` / `@pytest_asyncio.fixture` to their code. with this change - - `asyncio_mode=auto` is included in `pyproject.toml` making behavior consistent for all invocations of pytest - removes all redundant `@pytest_asyncio.fixture` and `@pytest.mark.asyncio` - for good measure, requires `pytest>=8.4` and `pytest-asyncio>=1.0` ## Test Plan - `./scripts/unit-tests.sh` - `uv run pytest tests/unit`	2025-07-11 13:00:24 -07:00
Francisco Arceo	6a6b66ae4f	chore: Adding unit tests for OpenAI vector stores and migrating SQLite-vec registry to kvstore (#2665) ... # What does this PR do? This PR refactors and the VectorIO backend logic for `sqlite-vec` and adds unit tests and fixtures to make it easy to test both `sqlite-vec` and `milvus`. Key changes: - `sqlite-vec` migrated to `kvstore` registry - added in-memory cache for sqlite-vec to be consistent with `milvus` - default fixtures moved to `conftest.py` - removed redundant tests from sqlite`-vec` - made `test_vector_io_openai_vector_stores.py` more easily extensible ## Test Plan Unit tests added testing inline providers. --------- Signed-off-by: Francisco Javier Arceo <farceo@redhat.com>	2025-07-10 14:22:13 -04:00
Francisco Arceo	83c89265e0	chore: Adding unit tests for Milvus and OpenAI compatibility (#2640) ... # What does this PR do? - Enabling Unit tests for Milvus to start to test OpenAI compatibility and fixing a few bugs. - Also fixed an inconsistency in the Milvus config between remote and inline. - Added pymilvus to extras for testing in CI I'm going to refactor this later to include the other inline providers so that we can catch issues sooner. I have another PR where I've been testing to find other bugs in the implementation (and required changes drafted here: https://github.com/meta-llama/llama-stack/pull/2617). ## 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.* --> --------- Signed-off-by: Francisco Javier Arceo <farceo@redhat.com>	2025-07-08 00:50:16 -07:00
Derek Higgins	f77d4d91f5	fix: handle encoding errors when adding files to vector store (#2574) ... - Add try-catch block around data.decode() to handle UnicodeDecodeError - Implement UTF-8 fallback when detected encoding fails - Return empty string when both encodings fail - add unit tests Fixes #2572: UnicodeDecodeError when uploading files with problematic encodings Signed-off-by: Derek Higgins <derekh@redhat.com>	2025-07-04 12:10:18 +02:00
Francisco Arceo	ea80ea63ac	chore: Updating chunk id generation to ensure uniqueness (#2618) ... # What does this PR do? This handles an edge case for `generate_chunk_id` if the concatenation of the `document_id` and `chunk_text` combination are not unique. Adding the window location ensures uniqueness. ## Test Plan Added unit test Signed-off-by: Francisco Javier Arceo <farceo@redhat.com>	2025-07-04 10:26:35 +05:30
Akram Ben Aissi	f4950f4ef0	fix: AccessDeniedError leads to HTTP 500 instead of error 403 (#2595) ... Resolves access control error visibility issues where 500 errors were returned instead of proper 403 responses with actionable error messages. • Enhance AccessDeniedError with detailed context and improve exception handling • Enhanced AccessDeniedError class to include user, action, and resource context - Added constructor parameters for action, resource, and user - Generate detailed error messages showing user principal, attributes, and attempted resource - Backward compatible with existing usage (falls back to generic message) • Updated exception handling in server.py - Import AccessDeniedError from access_control module - Return proper 403 status codes with detailed error messages - Separate handling for PermissionError (generic) vs AccessDeniedError (detailed) • Enhanced error context at raise sites - Updated routing_tables/common.py to pass action, resource, and user context - Updated agents persistence to include context in access denied errors - Provides better debugging information for access control issues • Added comprehensive unit tests - Created tests/unit/server/test_server.py with 13 test cases - Covers AccessDeniedError with and without context - Tests all exception types (ValidationError, BadRequestError, AuthenticationRequiredError, etc.) - Validates proper HTTP status codes and error message formats # 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 ``` server: port: 8321 access_policy: - permit: principal: admin actions: [create, read, delete] when: user with admin in groups - permit: actions: [read] when: user with system:authenticated in roles ``` then: ``` curl --request POST --url http://localhost:8321/v1/vector-dbs \ --header "Authorization: Bearer your-bearer" \ --data '{ "vector_db_id": "my_demo_vector_db", "embedding_model": "ibm-granite/granite-embedding-125m-english", "embedding_dimension": 768, "provider_id": "milvus" }' ``` depending if user is in group admin or not, you should get the `AccessDeniedError`. Before this PR, this was leading to an error 500 and `Traceback` displayed in the logs. After the PR, logs display a simpler error (unless DEBUG logging is set) and a 403 Forbidden error is returned on the HTTP side. --------- Signed-off-by: Akram Ben Aissi <<akram.benaissi@gmail.com>>	2025-07-03 10:50:49 -07:00
Sébastien Han	ac5fd57387	chore: remove nested imports (#2515) ... # What does this PR do? * Given that our API packages use "import *" in `__init.py__` we don't need to do `from llama_stack.apis.models.models` but simply from llama_stack.apis.models. The decision to use `import *` is debatable and should probably be revisited at one point. * Remove unneeded Ruff F401 rule * Consolidate Ruff F403 rule in the pyprojectfrom llama_stack.apis.models.models Signed-off-by: Sébastien Han <seb@redhat.com>	2025-06-26 08:01:05 +05:30
Ben Browning	2d9fd041eb	fix: annotations list and web_search_preview in Responses (#2520) ... # What does this PR do? These are a couple of fixes to get an example LangChain app working with our OpenAI Responses API implementation. The Responses API spec requires an annotations array in `output[*].content[*].annotations` and we were not providing one. So, this adds that as an empty list, even though we don't do anything to populate it yet. This prevents an error from client libraries like Langchain that expect this field to always exist, even if an empty list. The other fix is `web_search_preview` is a valid name for the web search tool in the Responses API, but we only responded to `web_search` or `web_search_preview_2025_03_11`. ## Test Plan The existing Responses unit tests were expanded to test these cases, via: ``` pytest -sv tests/unit/providers/agents/meta_reference/test_openai_responses.py ``` The existing test_openai_responses.py integration tests still pass with this change, tested as below with Fireworks: ``` uv run llama stack run llama_stack/templates/starter/run.yaml LLAMA_STACK_CONFIG=http://localhost:8321 \ uv run pytest -sv tests/integration/agents/test_openai_responses.py \ --text-model accounts/fireworks/models/llama4-scout-instruct-basic ``` Lastly, this example LangChain app now works with Llama stack (tested with Ollama in the starter template in this case). This LangChain code is using the example snippets for using Responses API at https://python.langchain.com/docs/integrations/chat/openai/#responses-api ```python from langchain_openai import ChatOpenAI llm = ChatOpenAI( base_url="http://localhost:8321/v1/openai/v1", api_key="fake", model="ollama/meta-llama/Llama-3.2-3B-Instruct", ) tool = {"type": "web_search_preview"} llm_with_tools = llm.bind_tools([tool]) response = llm_with_tools.invoke("What was a positive news story from today?") print(response.content) ``` Signed-off-by: Ben Browning <bbrownin@redhat.com>	2025-06-26 07:59:33 +05:30
Francisco Arceo	82f13fe83e	feat: Add ChunkMetadata to Chunk (#2497) ... # What does this PR do? Adding `ChunkMetadata` so we can properly delete embeddings later. More specifically, this PR refactors and extends the chunk metadata handling in the vector database and introduces a distinction between metadata used for model context and backend-only metadata required for chunk management, storage, and retrieval. It also improves chunk ID generation and propagation throughout the stack, enhances test coverage, and adds new utility modules. ```python class ChunkMetadata(BaseModel): """ `ChunkMetadata` is backend metadata for a `Chunk` that is used to store additional information about the chunk that will NOT be inserted into the context during inference, but is required for backend functionality. Use `metadata` in `Chunk` for metadata that will be used during inference. """ document_id: str | None = None chunk_id: str | None = None source: str | None = None created_timestamp: int | None = None updated_timestamp: int | None = None chunk_window: str | None = None chunk_tokenizer: str | None = None chunk_embedding_model: str | None = None chunk_embedding_dimension: int | None = None content_token_count: int | None = None metadata_token_count: int | None = None ``` Eventually we can migrate the document_id out of the `metadata` field. I've introduced the changes so that `ChunkMetadata` is backwards compatible with `metadata`. <!-- If resolving an issue, uncomment and update the line below --> Closes https://github.com/meta-llama/llama-stack/issues/2501 ## Test Plan Added unit tests --------- Signed-off-by: Francisco Javier Arceo <farceo@redhat.com>	2025-06-25 15:55:23 -04:00
ehhuang	d3b60507d7	feat: support auth attributes in inference/responses stores (#2389) ... # What does this PR do? Inference/Response stores now store user attributes when inserting, and respects them when fetching. ## Test Plan pytest tests/unit/utils/test_sqlstore.py	2025-06-20 10:24:45 -07:00
Ihar Hrachyshka	a2f054607d	fix: cancel scheduler tasks on shutdown (#2130) ... # What does this PR do? Scheduler: cancel tasks on shutdown. Otherwise the currently running tasks will never exit (before they actually complete), which means the process can't be properly shut down (only with SIGKILL). Ideally, we let tasks know that they are about to shutdown and give them some time to do so; but in the lack of the mechanism, it's better to cancel than linger forever. [//]: # (If resolving an issue, uncomment and update the line below) [//]: # (Closes #[issue-number]) ## Test Plan Start a long running task (e.g. torchtune or external kfp-provider training). Ctr-C the process in TTY. Confirm it exits in reasonable time. ``` ^CINFO: Shutting down INFO: Waiting for application shutdown. 13:32:26.187 - INFO - Shutting down 13:32:26.187 - INFO - Shutting down DatasetsRoutingTable 13:32:26.187 - INFO - Shutting down DatasetIORouter 13:32:26.187 - INFO - Shutting down TorchtuneKFPPostTrainingImpl Traceback (most recent call last): File "/opt/homebrew/Cellar/python@3.12/3.12.4/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/runners.py", line 118, in run return self._loop.run_until_complete(task) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/homebrew/Cellar/python@3.12/3.12.4/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/base_events.py", line 687, in run_until_complete return future.result() ^^^^^^^^^^^^^^^ asyncio.exceptions.CancelledError During handling of the above exception, another exception occurred: Traceback (most recent call last): File "<frozen runpy>", line 198, in _run_module_as_main File "<frozen runpy>", line 88, in _run_code File "/Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/dsl/executor_main.py", line 109, in <module> executor_main() File "/Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/dsl/executor_main.py", line 101, in executor_main output_file = executor.execute() ^^^^^^^^^^^^^^^^^^ File "/Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/dsl/executor.py", line 361, in execute result = self.func(**func_kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^ File "/var/folders/45/1q1rx6cn7jbcn2ty852w0g_r0000gn/T/tmp.RKpPrvTWDD/ephemeral_component.py", line 118, in component asyncio.run(recipe.setup()) File "/opt/homebrew/Cellar/python@3.12/3.12.4/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/runners.py", line 194, in run return runner.run(main) ^^^^^^^^^^^^^^^^ File "/opt/homebrew/Cellar/python@3.12/3.12.4/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/runners.py", line 123, in run raise KeyboardInterrupt() KeyboardInterrupt 13:32:31.219 - ERROR - Task 'component' finished with status FAILURE ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ INFO 2025-05-09 13:32:31,221 llama_stack.providers.utils.scheduler:221 scheduler: Job test-jobc3c2e1e4-859c-4852-a41d-ef29e55e3efa: Pipeline [1m[95m'test-jobc3c2e1e4-859c-4852-a41d-ef29e55e3efa'[1m[0m finished with status [1m[91mFAILURE[1m[0m. Inner task failed: [1m[96m'component'[1m[0m. ERROR 2025-05-09 13:32:31,223 llama_stack_provider_kfp_trainer.scheduler:54 scheduler: Job test-jobc3c2e1e4-859c-4852-a41d-ef29e55e3efa failed. ╭───────────────────────────────────── Traceback (most recent call last) ─────────────────────────────────────╮ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/src/llama_stack_provider_kfp_trainer/scheduler.py:45 │ │ in do │ │ │ │ 42 │ │ │ │ │ 43 │ │ │ job.status = JobStatus.running │ │ 44 │ │ │ try: │ │ ❱ 45 │ │ │ │ artifacts = self._to_artifacts(job.handler().output) │ │ 46 │ │ │ │ for artifact in artifacts: │ │ 47 │ │ │ │ │ on_artifact_collected_cb(artifact) │ │ 48 │ │ │ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/dsl/base_compon │ │ ent.py:101 in __call__ │ │ │ │ 98 │ │ │ │ f'{self.name}() missing {len(missing_arguments)} required ' │ │ 99 │ │ │ │ f'{argument_or_arguments}: {arguments}.') │ │ 100 │ │ │ │ ❱ 101 │ │ return pipeline_task.PipelineTask( │ │ 102 │ │ │ component_spec=self.component_spec, │ │ 103 │ │ │ args=task_inputs, │ │ 104 │ │ │ execute_locally=pipeline_context.Pipeline.get_default_pipeline() is │ │ │ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/dsl/pipeline_ta │ │ sk.py:187 in __init__ │ │ │ │ 184 │ │ ]) │ │ 185 │ │ │ │ 186 │ │ if execute_locally: │ │ ❱ 187 │ │ │ self._execute_locally(args=args) │ │ 188 │ │ │ 189 │ def _execute_locally(self, args: Dict[str, Any]) -> None: │ │ 190 │ │ """Execute the pipeline task locally. │ │ │ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/dsl/pipeline_ta │ │ sk.py:197 in _execute_locally │ │ │ │ 194 │ │ from kfp.local import task_dispatcher │ │ 195 │ │ │ │ 196 │ │ if self.pipeline_spec is not None: │ │ ❱ 197 │ │ │ self._outputs = pipeline_orchestrator.run_local_pipeline( │ │ 198 │ │ │ │ pipeline_spec=self.pipeline_spec, │ │ 199 │ │ │ │ arguments=args, │ │ 200 │ │ │ ) │ │ │ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/local/pipeline_ │ │ orchestrator.py:43 in run_local_pipeline │ │ │ │ 40 │ │ │ 41 │ # validate and access all global state in this function, not downstream │ │ 42 │ config.LocalExecutionConfig.validate() │ │ ❱ 43 │ return _run_local_pipeline_implementation( │ │ 44 │ │ pipeline_spec=pipeline_spec, │ │ 45 │ │ arguments=arguments, │ │ 46 │ │ raise_on_error=config.LocalExecutionConfig.instance.raise_on_error, │ │ │ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/local/pipeline_ │ │ orchestrator.py:108 in _run_local_pipeline_implementation │ │ │ │ 105 │ │ │ ) │ │ 106 │ │ return outputs │ │ 107 │ elif dag_status == status.Status.FAILURE: │ │ ❱ 108 │ │ log_and_maybe_raise_for_failure( │ │ 109 │ │ │ pipeline_name=pipeline_name, │ │ 110 │ │ │ fail_stack=fail_stack, │ │ 111 │ │ │ raise_on_error=raise_on_error, │ │ │ │ /Users/ihrachys/src/llama-stack-provider-kfp-trainer/.venv/lib/python3.12/site-packages/kfp/local/pipeline_ │ │ orchestrator.py:137 in log_and_maybe_raise_for_failure │ │ │ │ 134 │ │ logging_utils.format_task_name(task_name) for task_name in fail_stack) │ │ 135 │ msg = f'Pipeline {pipeline_name_with_color} finished with status │ │ {status_with_color}. Inner task failed: {task_chain_with_color}.' │ │ 136 │ if raise_on_error: │ │ ❱ 137 │ │ raise RuntimeError(msg) │ │ 138 │ with logging_utils.local_logger_context(): │ │ 139 │ │ logging.error(msg) │ │ 140 │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────╯ RuntimeError: Pipeline [1m[95m'test-jobc3c2e1e4-859c-4852-a41d-ef29e55e3efa'[1m[0m finished with status [1m[91mFAILURE[1m[0m. Inner task failed: [1m[96m'component'[1m[0m. INFO 2025-05-09 13:32:31,266 llama_stack.distribution.server.server:136 server: Shutting down DistributionInspectImpl INFO 2025-05-09 13:32:31,266 llama_stack.distribution.server.server:136 server: Shutting down ProviderImpl INFO: Application shutdown complete. INFO: Finished server process [26648] ``` [//]: # (## Documentation) Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>	2025-06-19 17:01:33 +02:00
ehhuang	db2cd9e8f3	feat: support filters in file search (#2472) ... # What does this PR do? Move to use vector_stores.search for file search tool in Responses, which supports filters. closes #2435 ## Test Plan Added e2e test with fitlers. myenv ❯ llama stack run llama_stack/templates/fireworks/run.yaml pytest -sv tests/verifications/openai_api/test_responses.py \ -k 'file_search and filters' \ --base-url=http://localhost:8321/v1/openai/v1 \ --model=meta-llama/Llama-3.3-70B-Instruct	2025-06-18 21:50:55 -07:00
Sumit Jaiswal	90d03552d4	feat: To add health check for faiss inline vector_io provider (#2319) ... # What does this PR do? <!-- Provide a short summary of what this PR does and why. Link to relevant issues if applicable. --> To add health check for faiss inline vector_io provider. I tried adding `async def health(self) -> HealthResponse:` like in inference provider, but it didn't worked for `inline->vector_io->faiss` provider. And via debug logs, I understood the critical issue, that the health responses are being stored with the API name as the key, not as a nested dictionary with provider IDs. This means that all providers of the same API type (e.g., "vector_io") will share the same health response, and only the last one processed will be visible in the API response. I've created a patch file that fixes this issue by: - Storing the original get_providers_health method - Creating a patched version that correctly maps health responses to providers - Applying the patch to the `ProviderImpl` class Not an expert, so please let me know, if there can be any other workaround using which I can get the health status updated directly from `faiss.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.* --> Added unit tests to test the provider patch implementation in the PR. Adding a screenshot with the FAISS inline vector_io health status as "OK" 	2025-06-18 17:56:25 +02:00
Varsha	2e8054bede	feat: Implement hybrid search in SQLite-vec (#2312) ... # What does this PR do? Add support for hybrid search mode in SQLite-vec provider, which combines keyword and vector search for better results. The implementation: - Adds hybrid search mode as a new option alongside vector and keyword search - Implements query_hybrid method in SQLiteVecIndex that: - First performs keyword search to get candidate matches - Then applies vector similarity search on those candidates - Updates documentation to reflect the new search mode This change improves search quality by leveraging both semantic similarity and keyword matching, while maintaining backward compatibility with existing vector and keyword search modes. ## Test Plan ``` pytest tests/unit/providers/vector_io/test_sqlite_vec.py -v -s --tb=short /Users/vnarsing/miniconda3/envs/stack-client/lib/python3.10/site-packages/pytest_asyncio/plugin.py:217: 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-8.3.5, pluggy-1.5.0 -- /Users/vnarsing/miniconda3/envs/stack-client/bin/python cachedir: .pytest_cache metadata: {'Python': '3.10.16', 'Platform': 'macOS-14.7.6-arm64-arm-64bit', 'Packages': {'pytest': '8.3.5', 'pluggy': '1.5.0'}, 'Plugins': {'html': '4.1.1', 'json-report': '1.5.0', 'timeout': '2.4.0', 'metadata': '3.1.1', 'anyio': '4.8.0', 'asyncio': '0.26.0', 'nbval': '0.11.0', 'cov': '6.1.1'}} rootdir: /Users/vnarsing/go/src/github/meta-llama/llama-stack configfile: pyproject.toml plugins: html-4.1.1, json-report-1.5.0, timeout-2.4.0, metadata-3.1.1, anyio-4.8.0, asyncio-0.26.0, nbval-0.11.0, cov-6.1.1 asyncio: mode=strict, asyncio_default_fixture_loop_scope=None, asyncio_default_test_loop_scope=function collected 10 items tests/unit/providers/vector_io/test_sqlite_vec.py::test_add_chunks PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_vector PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_full_text_search PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_hybrid PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_full_text_search_k_greater_than_results PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_chunk_id_conflict PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_generate_chunk_id PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_hybrid_no_keyword_matches PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_hybrid_score_threshold PASSED tests/unit/providers/vector_io/test_sqlite_vec.py::test_query_chunks_hybrid_different_embedding PASSED ``` --------- Signed-off-by: Varsha Prasad Narsing <varshaprasad96@gmail.com>	2025-06-13 15:54:06 -04:00
Ibrahim Haroon	a34cef925b	fix(faiss): handle case where distance is 0 by setting d to minimum positive… (#2387) ... # What does this PR do? Adds try-catch to faiss `query_vector` function for when the distance between the query embedding and an embedding within the vector db is 0 (identical vectors). Catches `ZeroDivisionError` and then appends `(1.0 / sys.float_info.min)` to `scores` to represent maximum similarity. <!-- If resolving an issue, uncomment and update the line below --> Closes [#2381] ## Test Plan Checkout this PR Execute this code and there will no longer be a `ZeroDivisionError` exception ``` from llama_stack_client import LlamaStackClient base_url = "http://localhost:8321" client = LlamaStackClient(base_url=base_url) models = client.models.list() embedding_model = ( em := next(m for m in models if m.model_type == "embedding") ).identifier embedding_dimension = 384 _ = client.vector_dbs.register( vector_db_id="foo_db", embedding_model=embedding_model, embedding_dimension=embedding_dimension, provider_id="faiss", ) chunk = { "content": "foo", "mime_type": "text/plain", "metadata": { "document_id": "foo-id" } } client.vector_io.insert(vector_db_id="foo_db", chunks=[chunk]) client.vector_io.query(vector_db_id="foo_db", query="foo") ``` ### Running unit tests `uv run pytest tests/unit/rag/test_rag_query.py -v` --------- Signed-off-by: Ben Browning <bbrownin@redhat.com> Co-authored-by: Ben Browning <bbrownin@redhat.com>	2025-06-07 16:09:46 -04:00
Sumit Jaiswal	33ecefd284	feat: To add health status check for remote VLLM (#2303) ... # What does this PR do? <!-- Provide a short summary of what this PR does and why. Link to relevant issues if applicable. --> To add health status check for remote VLLM <!-- 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.* --> PR includes the unit test to test the added health check implementation feature.	2025-06-06 15:33:12 -04:00
Ashwin Bharambe	3251b44d8a	refactor: unify stream and non-stream impls for responses (#2388) ... 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	2025-06-05 17:48:09 +02:00
grs	7c1998db25	feat: fine grained access control policy (#2264) ... 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>	2025-06-03 14:51:12 -07:00
Ben Browning	8bee2954be	feat: Structured output for Responses API (#2324) ... # 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>	2025-06-03 14:43:00 -07:00
Ashwin Bharambe	dbe4e84aca	feat(responses): implement full multi-turn support (#2295) ... 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 ```	2025-06-02 15:35:49 -07:00
Ben Browning	17f4414be9	fix: remote-vllm event loop blocking unit test on Mac (#2332) ... # What does this PR do? The remote-vllm `test_chat_completion_doesnt_block_event_loop` unit test was often failing for me on a Mac with a `httpx.ReadError`. I traced this back to the swap to the `AsyncOpenAI` client in the remote-vllm provider as where this started, and it looks like the async client needs a bit more accurate HTTP request handling from our mock server. So, this fixes that unit test to send proper Content-Type and Content-Length headers which makes the `AsyncOpenAI` client happier on Macs. ## Test Plan All the test_remote_vllm.py unit tests consistently pass for me on a Mac now, without any flaking in the event loop one. `pytest -s -v tests/unit/providers/inference/test_remote_vllm.py` Signed-off-by: Ben Browning <bbrownin@redhat.com>	2025-06-02 11:24:12 -04:00
Francisco Arceo	f328436831	feat: Enable ingestion of precomputed embeddings (#2317)	2025-05-31 04:03:37 -06:00
Ashwin Bharambe	bfdd15d1fa	fix(responses): use input, not original_input when storing the Response (#2300) ... 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	2025-05-28 13:17:48 -07:00
Ashwin Bharambe	5cdb29758a	feat(responses): add output_text delta events to responses (#2265) ... 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 ```	2025-05-27 13:07:14 -07:00
ehhuang	15b0a67555	feat: add responses input items api (#2239) ... # What does this PR do? TSIA ## Test Plan added integration and unit tests	2025-05-24 07:05:53 -07:00
ehhuang	5844c2da68	feat: add list responses API (#2233) ... # 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.	2025-05-23 13:16:48 -07:00
Varsha	e92301f2d7	feat(sqlite-vec): enable keyword search for sqlite-vec (#1439) ... # 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>	2025-05-21 15:24:24 -04:00
Derek Higgins	3339844fda	feat: Add "instructions" support to responses API (#2205) ... # 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>	2025-05-20 09:52:10 -07:00
Jash Gulabrai	1a770cf8ac	fix: Pass model parameter as config name to NeMo Customizer (#2218) ... # 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>	2025-05-20 09:51:39 -07:00
Ben Browning	10b1056dea	fix: multiple tool calls in remote-vllm chat_completion (#2161) ... # 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>	2025-05-15 11:23:29 -07:00
Ben Browning	b42eb1ccbc	fix: Responses API: handle type=None in streaming tool calls (#2166) ... # 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>	2025-05-14 14:16:33 -07:00
Ilya Kolchinsky	5052c3cbf3	fix: Fixed an "out of token budget" error when attempting a tool call via remote vLLM provider (#2114) ... # 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)	2025-05-14 13:11:02 -07:00
Ilya Kolchinsky	43d4447ff0	fix: remote vLLM tool execution now works when the last chunk contains the call arguments (#2112) ... # 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	2025-05-14 11:38:00 +02:00
Ben Browning	8e316c9b1e	feat: function tools in OpenAI Responses (#2094) ... # What does this PR do? This is a combination of what was previously 3 separate PRs - #2069, #2075, and #2083. It turns out all 3 of those are needed to land a working function calling Responses implementation. The web search builtin tool was already working, but this wires in support for custom function calling. I ended up combining all three into one PR because they all had lots of merge conflicts, both with each other but also with #1806 that just landed. And, because landing any of them individually would have only left a partially working implementation merged. The new things added here are: * Storing of input items from previous responses and restoring of those input items when adding previous responses to the conversation state * Handling of multiple input item messages roles, not just "user" messages. * Support for custom tools passed into the Responses API to enable function calling outside of just the builtin websearch tool. Closes #2074 Closes #2080 ## Test Plan ### Unit Tests Several new unit tests were added, and they all pass. Ran via: ``` python -m pytest -s -v tests/unit/providers/agents/meta_reference/test_openai_responses.py ``` ### Responses API Verification Tests I ran our verification run.yaml against multiple providers to ensure we were getting a decent pass rate. Specifically, I ensured the new custom tool verification test passed across multiple providers and that the multi-turn examples passed across at least some of the providers (some providers struggle with the multi-turn workflows still). Running the stack setup for verification testing: ``` llama stack run --image-type venv tests/verifications/openai-api-verification-run.yaml ``` Together, passing 100% as an example: ``` pytest -s -v 'tests/verifications/openai_api/test_responses.py' --provider=together-llama-stack ``` ## Documentation We will need to start documenting the OpenAI APIs, but for now the Responses stuff is still rapidly evolving so delaying that. --------- Signed-off-by: Derek Higgins <derekh@redhat.com> Signed-off-by: Ben Browning <bbrownin@redhat.com> Co-authored-by: Derek Higgins <derekh@redhat.com> Co-authored-by: Ashwin Bharambe <ashwin.bharambe@gmail.com>	2025-05-13 11:29:15 -07:00
Sébastien Han	c91e3552a3	feat: implementation for agent/session list and describe (#1606) ... Create a new agent: ``` curl --request POST \ --url http://localhost:8321/v1/agents \ --header 'Accept: application/json' \ --header 'Content-Type: application/json' \ --data '{ "agent_config": { "sampling_params": { "strategy": { "type": "greedy" }, "max_tokens": 0, "repetition_penalty": 1 }, "input_shields": [ "string" ], "output_shields": [ "string" ], "toolgroups": [ "string" ], "client_tools": [ { "name": "string", "description": "string", "parameters": [ { "name": "string", "parameter_type": "string", "description": "string", "required": true, "default": null } ], "metadata": { "property1": null, "property2": null } } ], "tool_choice": "auto", "tool_prompt_format": "json", "tool_config": { "tool_choice": "auto", "tool_prompt_format": "json", "system_message_behavior": "append" }, "max_infer_iters": 10, "model": "string", "instructions": "string", "enable_session_persistence": false, "response_format": { "type": "json_schema", "json_schema": { "property1": null, "property2": null } } } }' ``` Get agent: ``` curl http://127.0.0.1:8321/v1/agents/9abad4ab-2c77-45f9-9d16-46b79d2bea1f {"agent_id":"9abad4ab-2c77-45f9-9d16-46b79d2bea1f","agent_config":{"sampling_params":{"strategy":{"type":"greedy"},"max_tokens":0,"repetition_penalty":1.0},"input_shields":["string"],"output_shields":["string"],"toolgroups":["string"],"client_tools":[{"name":"string","description":"string","parameters":[{"name":"string","parameter_type":"string","description":"string","required":true,"default":null}],"metadata":{"property1":null,"property2":null}}],"tool_choice":"auto","tool_prompt_format":"json","tool_config":{"tool_choice":"auto","tool_prompt_format":"json","system_message_behavior":"append"},"max_infer_iters":10,"model":"string","instructions":"string","enable_session_persistence":false,"response_format":{"type":"json_schema","json_schema":{"property1":null,"property2":null}}},"created_at":"2025-03-12T16:18:28.369144Z"}% ``` List agents: ``` curl http://127.0.0.1:8321/v1/agents|jq % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 1680 100 1680 0 0 498k 0 --:--:-- --:--:-- --:--:-- 546k { "data": [ { "agent_id": "9abad4ab-2c77-45f9-9d16-46b79d2bea1f", "agent_config": { "sampling_params": { "strategy": { "type": "greedy" }, "max_tokens": 0, "repetition_penalty": 1.0 }, "input_shields": [ "string" ], "output_shields": [ "string" ], "toolgroups": [ "string" ], "client_tools": [ { "name": "string", "description": "string", "parameters": [ { "name": "string", "parameter_type": "string", "description": "string", "required": true, "default": null } ], "metadata": { "property1": null, "property2": null } } ], "tool_choice": "auto", "tool_prompt_format": "json", "tool_config": { "tool_choice": "auto", "tool_prompt_format": "json", "system_message_behavior": "append" }, "max_infer_iters": 10, "model": "string", "instructions": "string", "enable_session_persistence": false, "response_format": { "type": "json_schema", "json_schema": { "property1": null, "property2": null } } }, "created_at": "2025-03-12T16:18:28.369144Z" }, { "agent_id": "a6643aaa-96dd-46db-a405-333dc504b168", "agent_config": { "sampling_params": { "strategy": { "type": "greedy" }, "max_tokens": 0, "repetition_penalty": 1.0 }, "input_shields": [ "string" ], "output_shields": [ "string" ], "toolgroups": [ "string" ], "client_tools": [ { "name": "string", "description": "string", "parameters": [ { "name": "string", "parameter_type": "string", "description": "string", "required": true, "default": null } ], "metadata": { "property1": null, "property2": null } } ], "tool_choice": "auto", "tool_prompt_format": "json", "tool_config": { "tool_choice": "auto", "tool_prompt_format": "json", "system_message_behavior": "append" }, "max_infer_iters": 10, "model": "string", "instructions": "string", "enable_session_persistence": false, "response_format": { "type": "json_schema", "json_schema": { "property1": null, "property2": null } } }, "created_at": "2025-03-12T16:17:12.811273Z" } ] } ``` Create sessions: ``` curl --request POST \ --url http://localhost:8321/v1/agents/{agent_id}/session \ --header 'Accept: application/json' \ --header 'Content-Type: application/json' \ --data '{ "session_name": "string" }' ``` List sessions: ``` curl http://127.0.0.1:8321/v1/agents/9abad4ab-2c77-45f9-9d16-46b79d2bea1f/sessions|jq % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 263 100 263 0 0 90099 0 --:--:-- --:--:-- --:--:-- 128k [ { "session_id": "2b15c4fc-e348-46c1-ae32-f6d424441ac1", "session_name": "string", "turns": [], "started_at": "2025-03-12T17:19:17.784328" }, { "session_id": "9432472d-d483-4b73-b682-7b1d35d64111", "session_name": "string", "turns": [], "started_at": "2025-03-12T17:19:19.885834" } ] ``` Signed-off-by: Sébastien Han <seb@redhat.com>	2025-05-07 14:49:23 +02:00
Kevin Postlethwait	a57985eeac	fix: add check for interleavedContent (#1973) ... # What does this PR do? Checks for RAGDocument of type InterleavedContent I noticed when stepping through the code that the supported types for `RAGDocument` included `InterleavedContent` as a content type. This type is not checked against before putting the `doc.content` is regex matched against. This would cause a runtime error. This change adds an explicit check for type. The only other part that I'm unclear on is how to handle the `ImageContent` type since this would always just return `<image>` which seems like an undesired behavior. Should the `InterleavedContent` type be removed from `RAGDocument` and replaced with `URI | str`? ## Test Plan [//]: # (## Documentation) --------- Signed-off-by: Kevin <kpostlet@redhat.com>	2025-05-06 09:55:07 -07:00
Derek Higgins	2e807b38cc	chore: Add fixtures to conftest.py (#2067) ... Add fixtures for SqliteKVStore, DiskDistributionRegistry and CachedDiskDistributionRegistry. And use them in tests that had all been duplicating similar setups. ## Test Plan unit tests continue to run Signed-off-by: Derek Higgins <derekh@redhat.com>	2025-05-06 13:57:48 +02:00
Ben Browning	f1b103e6c8	fix: openai_compat messages system/assistant non-str content (#2095) ... # What does this PR do? When converting OpenAI message content for the "system" and "assistant" roles to Llama Stack inference APIs (used for some providers when dealing with Llama models via OpenAI API requests to get proper prompt / tool handling), we were not properly converting any non-string content. I discovered this while running the new Responses AI verification suite against the Fireworks provider, but instead of fixing it as part of some ongoing work there split this out into a separate PR. This fixes that, by using the `openai_content_to_content` helper we used elsewhere to ensure content parts were mapped properly. ## Test Plan I added a couple of new tests to `test_openai_compat` to reproduce this issue and validate its fix. I ran those as below: ``` python -m pytest -s -v tests/unit/providers/utils/inference/test_openai_compat.py ``` Signed-off-by: Ben Browning <bbrownin@redhat.com>	2025-05-02 13:09:27 -07:00
Ihar Hrachyshka	9e6561a1ec	chore: enable pyupgrade fixes (#1806) ... # What does this PR do? The goal of this PR is code base modernization. Schema reflection code needed a minor adjustment to handle UnionTypes and collections.abc.AsyncIterator. (Both are preferred for latest Python releases.) Note to reviewers: almost all changes here are automatically generated by pyupgrade. Some additional unused imports were cleaned up. The only change worth of note can be found under `docs/openapi_generator` and `llama_stack/strong_typing/schema.py` where reflection code was updated to deal with "newer" types. Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>	2025-05-01 14:23:50 -07:00
Matthew Farrellee	88a796ca5a	fix: allow use of models registered at runtime (#1980) ... # What does this PR do? fix a bug where models registered at runtime could not be used. ``` $ llama-stack-client models register test-model --provider-id nvidia --provider-model-id meta/llama-3.1-70b-instruct $ curl http://localhost:8321/v1/openai/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "test-model", "messages": [{"role": "user", "content": "What is the weather like in Boston today?"}] }' =(client)=> {"detail":"Internal server error: An unexpected error occurred."} =(server)=> TypeError: Missing required arguments; Expected either ('messages' and 'model') or ('messages', 'model' and 'stream') arguments to be given ``` *root cause:* test-model is not added to ModelRegistryHelper's alias_to_provider_id_map. as part of the fix, this adds tests for ModelRegistryHelper and defines its expected behavior. user visible behavior changes - | action | existing behavior | new behavior | | -- | -- | -- | | double register | success (but no change) | error | | register unknown | success (fail when used) | error | existing behavior for register unknown model and double register - ``` $ llama-stack-client models register test-model --provider-id nvidia --provider-model-id meta/llama-3.1-70b-instruct-unknown Successfully registered model test-model $ llama-stack-client models list | grep test-model │ llm │ test-model │ meta/llama-3.1-70b-instruct-unknown │ │ nv… │ $ llama-stack-client models register test-model --provider-id nvidia --provider-model-id meta/llama-3.1-70b-instruct Successfully registered model test-model $ llama-stack-client models list | grep test-model │ llm │ test-model │ meta/llama-3.1-70b-instruct-unknown │ │ nv… │ ``` new behavior for register unknown - ``` $ llama-stack-client models register test-model --provider-id nvidia --provider-model-id meta/llama-3.1-70b-instruct-unknown ╭──────────────────────────────────────────────────────────────────────────────────────────────────╮ │ Failed to register model │ │ │ │ Error Type: BadRequestError │ │ Details: Error code: 400 - {'detail': "Invalid value: Model id │ │ 'meta/llama-3.1-70b-instruct-unknown' is not supported. Supported ids are: │ │ meta/llama-3.1-70b-instruct, snowflake/arctic-embed-l, meta/llama-3.2-1b-instruct, │ │ nvidia/nv-embedqa-mistral-7b-v2, meta/llama-3.2-90b-vision-instruct, meta/llama-3.2-3b-instruct, │ │ meta/llama-3.2-11b-vision-instruct, meta/llama-3.1-405b-instruct, meta/llama3-8b-instruct, │ │ meta/llama3-70b-instruct, nvidia/llama-3.2-nv-embedqa-1b-v2, meta/llama-3.1-8b-instruct, │ │ nvidia/nv-embedqa-e5-v5"} │ ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯ ``` new behavior for double register - ``` $ llama-stack-client models register test-model --provider-id nvidia --provider-model-id meta/llama-3.1-70b-instruct Successfully registered model test-model $ llama-stack-client models register test-model --provider-id nvidia --provider-model-id meta/llama-3.2-1b-instruct ╭──────────────────────────────────────────────────────────────────────────────────────────────────╮ │ Failed to register model │ │ │ │ Error Type: BadRequestError │ │ Details: Error code: 400 - {'detail': "Invalid value: Model id 'test-model' is already │ │ registered. Please use a different id or unregister it first."} │ ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯ ``` ## Test Plan ``` uv run pytest -v tests/unit/providers/utils/test_model_registry.py ```	2025-05-01 12:00:58 -07:00
Derek Higgins	64829947d0	feat: Add temperature support to responses API (#2065) ... # What does this PR do? Add support for the temperature to the responses API ## Test Plan Manually tested simple case unit tests added for simple case and tool calls Signed-off-by: Derek Higgins <derekh@redhat.com>	2025-05-01 11:47:58 -07:00
Ben Browning	6378c2a2f3	fix: resolve BuiltinTools to strings for vllm tool_call messages (#2071) ... # What does this PR do? When the result of a ToolCall gets passed back into vLLM for the model to handle the tool call result (as is often the case in agentic tool-calling workflows), we forgot to handle the case where BuiltinTool calls are not string values but instead instances of the BuiltinTool enum. This fixes that, properly converting those enums to string values before trying to serialize them into an OpenAI chat completion request to vLLM. PR #1931 fixed a bug where we weren't passing these tool calling results back into vLLM, but as a side-effect it created this serialization bug when using BuiltinTools. Closes #2070 ## Test Plan I added a new unit test to the openai_compat unit tests to cover this scenario, ensured the new test failed before this fix, and all the existing tests there plus the new one passed with this fix. ``` python -m pytest -s -v tests/unit/providers/utils/inference/test_openai_compat.py ``` Signed-off-by: Ben Browning <bbrownin@redhat.com>	2025-05-01 08:47:29 -04:00