Sambanova inference provider (#555)

# What does this PR do?

This PR adds SambaNova as one of the Provider

- Add SambaNova as a provider

## Test Plan
Test the functional command
```
pytest -s -v --providers inference=sambanova llama_stack/providers/tests/inference/test_embeddings.py llama_stack/providers/tests/inference/test_prompt_adapter.py llama_stack/providers/tests/inference/test_text_inference.py llama_stack/providers/tests/inference/test_vision_inference.py --env SAMBANOVA_API_KEY=<sambanova-api-key>
```

Test the distribution template:
```
# Docker
LLAMA_STACK_PORT=5001
docker run -it -p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
  llamastack/distribution-sambanova \
  --port $LLAMA_STACK_PORT \
  --env SAMBANOVA_API_KEY=$SAMBANOVA_API_KEY

# Conda
llama stack build --template sambanova --image-type conda
llama stack run ./run.yaml \
  --port $LLAMA_STACK_PORT \
  --env SAMBANOVA_API_KEY=$SAMBANOVA_API_KEY
```

## Source
[SambaNova API Documentation](https://cloud.sambanova.ai/apis)

## Before submitting

- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [Y] Ran pre-commit to handle lint / formatting issues.
- [Y] Read the [contributor
guideline](https://github.com/meta-llama/llama-stack/blob/main/CONTRIBUTING.md),
      Pull Request section?
- [Y] Updated relevant documentation.
- [Y ] Wrote necessary unit or integration tests.

---------

Co-authored-by: Ashwin Bharambe <ashwin.bharambe@gmail.com>

llama_stack/providers/tests/inference/fixtures.py

@@ -23,6 +23,7 @@ from llama_stack.providers.remote.inference.fireworks import FireworksImplConfig
 from llama_stack.providers.remote.inference.groq import GroqConfig
 from llama_stack.providers.remote.inference.nvidia import NVIDIAConfig
 from llama_stack.providers.remote.inference.ollama import OllamaImplConfig
+from llama_stack.providers.remote.inference.sambanova import SambaNovaImplConfig
 from llama_stack.providers.remote.inference.tgi import TGIImplConfig
 from llama_stack.providers.remote.inference.together import TogetherImplConfig
 from llama_stack.providers.remote.inference.vllm import VLLMInferenceAdapterConfig
@@ -232,6 +233,23 @@ def inference_tgi() -> ProviderFixture:
 
 
 @pytest.fixture(scope="session")
+def inference_sambanova() -> ProviderFixture:
+    return ProviderFixture(
+        providers=[
+            Provider(
+                provider_id="sambanova",
+                provider_type="remote::sambanova",
+                config=SambaNovaImplConfig(
+                    api_key=get_env_or_fail("SAMBANOVA_API_KEY"),
+                ).model_dump(),
+            )
+        ],
+        provider_data=dict(
+            sambanova_api_key=get_env_or_fail("SAMBANOVA_API_KEY"),
+        ),
+    )
+
+
 def inference_sentence_transformers() -> ProviderFixture:
     return ProviderFixture(
         providers=[
@@ -282,6 +300,7 @@ INFERENCE_FIXTURES = [
     "cerebras",
     "nvidia",
     "tgi",
+    "sambanova",
 ]
 
 

llama_stack/providers/tests/inference/test_model_registration.py

@@ -59,7 +59,7 @@ class TestModelRegistration:
             },
         )
 
-        with pytest.raises(AssertionError) as exc_info:
+        with pytest.raises(ValueError) as exc_info:
             await models_impl.register_model(
                 model_id="custom-model-2",
                 metadata={

llama_stack/providers/tests/inference/test_text_inference.py

@@ -385,6 +385,12 @@ class TestInference:
             # TODO(aidand): Remove this skip once Groq's tool calling for Llama3.2 works better
             pytest.skip("Groq's tool calling for Llama3.2 doesn't work very well")
 
+        if provider.__provider_spec__.provider_type == "remote::sambanova" and (
+            "-1B-" in inference_model or "-3B-" in inference_model
+        ):
+            # TODO(snova-edawrdm): Remove this skip once SambaNova's tool calling for 1B/ 3B
+            pytest.skip("Sambanova's tool calling for lightweight models don't work")
+
         messages = sample_messages + [
             UserMessage(
                 content="What's the weather like in San Francisco?",
@@ -431,6 +437,9 @@ class TestInference:
         ):
             # TODO(aidand): Remove this skip once Groq's tool calling for Llama3.2 works better
             pytest.skip("Groq's tool calling for Llama3.2 doesn't work very well")
+        if provider.__provider_spec__.provider_type == "remote::sambanova":
+            # TODO(snova-edawrdm): Remove this skip once SambaNova's tool calling under streaming is supported (we are working on it)
+            pytest.skip("Sambanova's tool calling for streaming doesn't work")
 
         messages = sample_messages + [
             UserMessage(

llama_stack/providers/tests/inference/test_vision_inference.py

@@ -59,6 +59,7 @@ class TestVisionModelInference:
             "remote::fireworks",
             "remote::ollama",
             "remote::vllm",
+            "remote::sambanova",
         ):
             pytest.skip(
                 "Other inference providers don't support vision chat completion() yet"
@@ -98,6 +99,7 @@ class TestVisionModelInference:
             "remote::fireworks",
             "remote::ollama",
             "remote::vllm",
+            "remote::sambanova",
         ):
             pytest.skip(
                 "Other inference providers don't support vision chat completion() yet"