Merge branch 'main' of https://github.com/meta-llama/llama-stack into register_custom_model

llama_stack/providers/remote/inference/nvidia/NVIDIA.md

@@ -0,0 +1,85 @@
+# NVIDIA Inference Provider for LlamaStack
+
+This provider enables running inference using NVIDIA NIM.
+
+## Features
+- Endpoints for completions, chat completions, and embeddings for registered models
+
+## Getting Started
+
+### Prerequisites
+
+- LlamaStack with NVIDIA configuration
+- Access to NVIDIA NIM deployment
+- NIM for model to use for inference is deployed
+
+### Setup
+
+Build the NVIDIA environment:
+
+```bash
+llama stack build --template nvidia --image-type conda
+```
+
+### Basic Usage using the LlamaStack Python Client
+
+#### Initialize the client
+
+```python
+import os
+
+os.environ["NVIDIA_API_KEY"] = (
+    ""  # Required if using hosted NIM endpoint. If self-hosted, not required.
+)
+os.environ["NVIDIA_BASE_URL"] = "http://nim.test"  # NIM URL
+
+from llama_stack.distribution.library_client import LlamaStackAsLibraryClient
+
+client = LlamaStackAsLibraryClient("nvidia")
+client.initialize()
+```
+
+### Create Completion
+
+```python
+response = client.completion(
+    model_id="meta-llama/Llama-3.1-8b-Instruct",
+    content="Complete the sentence using one word: Roses are red, violets are :",
+    stream=False,
+    sampling_params={
+        "max_tokens": 50,
+    },
+)
+print(f"Response: {response.content}")
+```
+
+### Create Chat Completion
+
+```python
+response = client.chat_completion(
+    model_id="meta-llama/Llama-3.1-8b-Instruct",
+    messages=[
+        {
+            "role": "system",
+            "content": "You must respond to each message with only one word",
+        },
+        {
+            "role": "user",
+            "content": "Complete the sentence using one word: Roses are red, violets are:",
+        },
+    ],
+    stream=False,
+    sampling_params={
+        "max_tokens": 50,
+    },
+)
+print(f"Response: {response.completion_message.content}")
+```
+
+### Create Embeddings
+```python
+response = client.embeddings(
+    model_id="meta-llama/Llama-3.1-8b-Instruct", contents=["foo", "bar", "baz"]
+)
+print(f"Embeddings: {response.embeddings}")
+```

llama_stack/providers/remote/inference/nvidia/models.py

@@ -48,6 +48,10 @@ MODEL_ENTRIES = [
         "meta/llama-3.2-90b-vision-instruct",
         CoreModelId.llama3_2_90b_vision_instruct.value,
     ),
+    build_hf_repo_model_entry(
+        "meta/llama-3.3-70b-instruct",
+        CoreModelId.llama3_3_70b_instruct.value,
+    ),
     # NeMo Retriever Text Embedding models -
     #
     # https://docs.nvidia.com/nim/nemo-retriever/text-embedding/latest/support-matrix.html

llama_stack/providers/remote/inference/nvidia/nvidia.py

@@ -129,6 +129,14 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
             base_url = special_model_urls[provider_model_id]
         return _get_client_for_base_url(base_url)
 
+    async def _get_provider_model_id(self, model_id: str) -> str:
+        if not self.model_store:
+            raise RuntimeError("Model store is not set")
+        model = await self.model_store.get_model(model_id)
+        if model is None:
+            raise ValueError(f"Model {model_id} is unknown")
+        return model.provider_model_id
+
     async def completion(
         self,
         model_id: str,
@@ -147,7 +155,7 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
         # removing this health check as NeMo customizer endpoint health check is returning 404
         # await check_health(self._config)  # this raises errors
 
-        provider_model_id = self.get_provider_model_id(model_id)
+        provider_model_id = await self._get_provider_model_id(model_id)
         request = convert_completion_request(
             request=CompletionRequest(
                 model=provider_model_id,
@@ -191,7 +199,7 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
         #
         flat_contents = [content.text if isinstance(content, TextContentItem) else content for content in contents]
         input = [content.text if isinstance(content, TextContentItem) else content for content in flat_contents]
-        model = self.get_provider_model_id(model_id)
+        provider_model_id = await self._get_provider_model_id(model_id)
 
         extra_body = {}
 
@@ -214,8 +222,8 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
             extra_body["input_type"] = task_type_options[task_type]
 
         try:
-            response = await self._get_client(model).embeddings.create(
-                model=model,
+            response = await self._get_client(provider_model_id).embeddings.create(
+                model=provider_model_id,
                 input=input,
                 extra_body=extra_body,
             )
@@ -249,10 +257,10 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
 
         # await check_health(self._config)  # this raises errors
 
-        provider_model_id = self.get_provider_model_id(model_id)
+        provider_model_id = await self._get_provider_model_id(model_id)
         request = await convert_chat_completion_request(
             request=ChatCompletionRequest(
-                model=self.get_provider_model_id(model_id),
+                model=provider_model_id,
                 messages=messages,
                 sampling_params=sampling_params,
                 response_format=response_format,
@@ -297,7 +305,7 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
         guided_choice: Optional[List[str]] = None,
         prompt_logprobs: Optional[int] = None,
     ) -> OpenAICompletion:
-        provider_model_id = self.get_provider_model_id(model)
+        provider_model_id = await self._get_provider_model_id(model)
 
         params = await prepare_openai_completion_params(
             model=provider_model_id,
@@ -350,7 +358,7 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
         top_p: Optional[float] = None,
         user: Optional[str] = None,
     ) -> Union[OpenAIChatCompletion, AsyncIterator[OpenAIChatCompletionChunk]]:
-        provider_model_id = self.get_provider_model_id(model)
+        provider_model_id = await self._get_provider_model_id(model)
 
         params = await prepare_openai_completion_params(
             model=provider_model_id,