add nvidia distribution (#565)

# What does this PR do?

adds nvidia template for creating a distribution using inference adapter
for NVIDIA NIMs.

## Test Plan

Please describe:
Build llama stack distribution for nvidia using the template, docker and
conda.
```bash
(.venv) local-cdgamarose@a4u8g-0006:~/llama-stack$ llama-stack-client configure --endpoint http://localhost:5000
Done! You can now use the Llama Stack Client CLI with endpoint http://localhost:5000
(.venv) local-cdgamarose@a4u8g-0006:~/llama-stack$ llama-stack-client models list
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ identifier                       ┃ provider_id ┃ provider_resource_id       ┃ metadata ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Llama3.1-8B-Instruct             │ nvidia      │ meta/llama-3.1-8b-instruct │ {}       │
│ meta-llama/Llama-3.2-3B-Instruct │ nvidia      │ meta/llama-3.2-3b-instruct │ {}       │
└──────────────────────────────────┴─────────────┴────────────────────────────┴──────────┘
(.venv) local-cdgamarose@a4u8g-0006:~/llama-stack$ llama-stack-client inference chat-completion --message "hello, write me a 2 sentence poem"
ChatCompletionResponse(
    completion_message=CompletionMessage(
        content='Here is a 2 sentence poem:\n\nThe sun sets slow and paints the sky, \nA gentle hue of pink that makes me sigh.',
        role='assistant',
        stop_reason='end_of_turn',
        tool_calls=[]
    ),
    logprobs=None
)
```

## Before submitting

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

---------

Co-authored-by: Matthew Farrellee <matt@cs.wisc.edu>

distributions/inline-nvidia/build.yaml

@@ -0,0 +1 @@
+../../llama_stack/templates/nvidia/build.yaml

distributions/inline-nvidia/compose.yaml

@@ -0,0 +1,58 @@
+services:
+  nim:
+    image: ${DOCKER_IMAGE:-nvcr.io/nim/meta/llama-3.1-8b-instruct:latest}
+    network_mode: "host"
+    volumes:
+    - nim-llm-cache:/opt/nim/.cache
+    ports:
+      - "8000:8000"
+    shm_size: 16G
+    environment:
+      - CUDA_VISIBLE_DEVICES=0
+      - NIM_HTTP_API_PORT=8000
+      - NIM_TRITON_LOG_VERBOSE=1
+      - NGC_API_KEY=${NIM_NGC_API_KEY:-${NGC_API_KEY:-ngcapikey}}
+    command: []
+    deploy:
+      resources:
+        reservations:
+          devices:
+          - driver: nvidia
+            # that's the closest analogue to --gpus; provide
+            # an integer amount of devices or 'all'
+            count: 1
+            # Devices are reserved using a list of capabilities, making
+            # capabilities the only required field. A device MUST
+            # satisfy all the requested capabilities for a successful
+            # reservation.
+            capabilities: [gpu]
+    runtime: nvidia
+    healthcheck:
+      test: ["CMD", "curl", "http://localhost:8000/v1/health/ready"]
+      interval: 5s
+      timeout: 5s
+      retries: 30
+      start_period: 120s
+  llamastack:
+    depends_on:
+    - nim
+    image: distribution-nvidia:dev
+    network_mode: "host"
+    volumes:
+      - ~/.llama:/root/.llama
+      - ./run.yaml:/root/llamastack-run-nvidia.yaml
+    ports:
+      - "5000:5000"
+    environment:
+      - INFERENCE_MODEL=${INFERENCE_MODEL:-Llama3.1-8B-Instruct}
+      - NVIDIA_API_KEY=${NVIDIA_API_KEY:-}
+    entrypoint: bash -c "python -m llama_stack.distribution.server.server --yaml-config /root/llamastack-run-nvidia.yaml"
+    deploy:
+      restart_policy:
+        condition: on-failure
+        delay: 3s
+        max_attempts: 5
+        window: 60s
+volumes:
+  nim-llm-cache:
+    driver: local

distributions/inline-nvidia/run.yaml

@@ -0,0 +1,100 @@
+version: '2'
+image_name: nvidia
+conda_env: nvidia
+apis:
+- agents
+- datasetio
+- eval
+- inference
+- memory
+- safety
+- scoring
+- telemetry
+- tool_runtime
+providers:
+  inference:
+  - provider_id: nvidia
+    provider_type: remote::nvidia
+    config:
+      url: http://localhost:8000
+      api_key: ${env.NVIDIA_API_KEY} # TODO: don't need api key, code adjustments needed
+  memory:
+  - provider_id: faiss
+    provider_type: inline::faiss
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/faiss_store.db
+  safety:
+  - provider_id: llama-guard
+    provider_type: inline::llama-guard
+    config: {}
+  agents:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      persistence_store:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/agents_store.db
+  telemetry:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
+      sinks: ${env.TELEMETRY_SINKS:console,sqlite}
+      sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/nvidia/trace_store.db}
+  eval:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config: {}
+  datasetio:
+  - provider_id: huggingface
+    provider_type: remote::huggingface
+    config: {}
+  - provider_id: localfs
+    provider_type: inline::localfs
+    config: {}
+  scoring:
+  - provider_id: basic
+    provider_type: inline::basic
+    config: {}
+  - provider_id: llm-as-judge
+    provider_type: inline::llm-as-judge
+    config: {}
+  - provider_id: braintrust
+    provider_type: inline::braintrust
+    config:
+      openai_api_key: ${env.OPENAI_API_KEY:}
+  tool_runtime:
+  - provider_id: brave-search
+    provider_type: remote::brave-search
+    config:
+      api_key: ${env.BRAVE_SEARCH_API_KEY:}
+      max_results: 3
+  - provider_id: tavily-search
+    provider_type: remote::tavily-search
+    config:
+      api_key: ${env.TAVILY_SEARCH_API_KEY:}
+      max_results: 3
+  - provider_id: code-interpreter
+    provider_type: inline::code-interpreter
+    config: {}
+  - provider_id: memory-runtime
+    provider_type: inline::memory-runtime
+    config: {}
+metadata_store:
+  type: sqlite
+  db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/registry.db
+models:
+- metadata: {}
+  model_id: ${env.INFERENCE_MODEL}
+  provider_id: nvidia
+  model_type: llm
+shields: []
+memory_banks: []
+datasets: []
+scoring_fns: []
+eval_tasks: []
+tool_groups: []

distributions/remote-nvidia/build.yaml

@@ -0,0 +1 @@
+../../llama_stack/templates/nvidia/build.yaml

distributions/remote-nvidia/compose.yaml

@@ -0,0 +1,19 @@
+services:
+  llamastack:
+    image: distribution-nvidia:dev
+    network_mode: "host"
+    volumes:
+      - ~/.llama:/root/.llama
+      - ./run.yaml:/root/llamastack-run-nvidia.yaml
+    ports:
+      - "5000:5000"
+    environment:
+      - INFERENCE_MODEL=${INFERENCE_MODEL:-Llama3.1-8B-Instruct}
+      - NVIDIA_API_KEY=${NVIDIA_API_KEY:-}
+    entrypoint: bash -c "python -m llama_stack.distribution.server.server --yaml-config /root/llamastack-run-nvidia.yaml"
+    deploy:
+      restart_policy:
+        condition: on-failure
+        delay: 3s
+        max_attempts: 5
+        window: 60s

distributions/remote-nvidia/run.yaml

@@ -0,0 +1 @@
+../../llama_stack/templates/nvidia/run.yaml

docs/source/distributions/index.md

@@ -20,6 +20,7 @@ If so, we suggest:
   - {dockerhub}`distribution-remote-vllm` ([Guide](self_hosted_distro/remote-vllm))
   - {dockerhub}`distribution-meta-reference-gpu` ([Guide](self_hosted_distro/meta-reference-gpu))
   - {dockerhub}`distribution-tgi` ([Guide](self_hosted_distro/tgi))
+  - {dockerhub} `distribution-nvidia` ([Guide](self_hosted_distro/nvidia))
 
 - **Are you running on a "regular" desktop machine?**
 If so, we suggest:

docs/source/distributions/remote_hosted_distro/nvidia.md

@@ -0,0 +1,65 @@
+# NVIDIA Distribution
+
+The `llamastack/distribution-nvidia` distribution consists of the following provider configurations.
+
+| API | Provider(s) |
+|-----|-------------|
+| agents | `inline::meta-reference` |
+| datasetio | `remote::huggingface`, `inline::localfs` |
+| eval | `inline::meta-reference` |
+| inference | `remote::nvidia` |
+| memory | `inline::faiss` |
+| safety | `inline::llama-guard` |
+| scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
+| telemetry | `inline::meta-reference` |
+| tool_runtime | `remote::brave-search`, `remote::tavily-search`, `inline::code-interpreter`, `inline::memory-runtime` |
+
+
+### Environment Variables
+
+The following environment variables can be configured:
+
+- `LLAMASTACK_PORT`: Port for the Llama Stack distribution server (default: `5001`)
+- `NVIDIA_API_KEY`: NVIDIA API Key (default: ``)
+
+### Models
+
+The following models are available by default:
+
+- `${env.INFERENCE_MODEL} (None)`
+
+
+### Prerequisite: API Keys
+
+Make sure you have access to a NVIDIA API Key. You can get one by visiting [https://build.nvidia.com/](https://build.nvidia.com/).
+
+
+## Running Llama Stack with NVIDIA
+
+You can do this via Conda (build code) or Docker which has a pre-built image.
+
+### Via Docker
+
+This method allows you to get started quickly without having to build the distribution code.
+
+```bash
+LLAMA_STACK_PORT=5001
+docker run \
+  -it \
+  -p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
+  -v ./run.yaml:/root/my-run.yaml \
+  llamastack/distribution-nvidia \
+  --yaml-config /root/my-run.yaml \
+  --port $LLAMA_STACK_PORT \
+  --env NVIDIA_API_KEY=$NVIDIA_API_KEY
+```
+
+### Via Conda
+
+```bash
+llama stack build --template nvidia --image-type conda
+llama stack run ./run.yaml \
+  --port 5001 \
+  --env NVIDIA_API_KEY=$NVIDIA_API_KEY
+  --env INFERENCE=$INFERENCE_MODEL
+```

docs/source/distributions/self_hosted_distro/nvidia.md

@@ -0,0 +1,60 @@
+# NVIDIA Distribution
+
+The `llamastack/distribution-nvidia` distribution consists of the following provider configurations.
+
+| API | Provider(s) |
+|-----|-------------|
+| agents | `inline::meta-reference` |
+| inference | `remote::nvidia` |
+| memory | `inline::faiss`, `remote::chromadb`, `remote::pgvector` |
+| safety | `inline::llama-guard` |
+| telemetry | `inline::meta-reference` |
+
+
+### Environment Variables
+
+The following environment variables can be configured:
+
+- `LLAMASTACK_PORT`: Port for the Llama Stack distribution server (default: `5001`)
+- `NVIDIA_API_KEY`: NVIDIA API Key (default: ``)
+
+### Models
+
+The following models are available by default:
+
+- `${env.INFERENCE_MODEL} (None)`
+
+
+### Prerequisite: API Keys
+
+Make sure you have access to a NVIDIA API Key. You can get one by visiting [https://build.nvidia.com/](https://build.nvidia.com/).
+
+
+## Running Llama Stack with NVIDIA
+
+You can do this via Conda (build code) or Docker which has a pre-built image.
+
+### Via Docker
+
+This method allows you to get started quickly without having to build the distribution code.
+
+```bash
+LLAMA_STACK_PORT=5001
+docker run \
+  -it \
+  -p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
+  -v ./run.yaml:/root/my-run.yaml \
+  llamastack/distribution-nvidia \
+  --yaml-config /root/my-run.yaml \
+  --port $LLAMA_STACK_PORT \
+  --env NVIDIA_API_KEY=$NVIDIA_API_KEY
+```
+
+### Via Conda
+
+```bash
+llama stack build --template nvidia --image-type conda
+llama stack run ./run.yaml \
+  --port 5001 \
+  --env NVIDIA_API_KEY=$NVIDIA_API_KEY
+```

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

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 import os
-from typing import Optional
+from typing import Any, Dict, Optional
 
 from llama_models.schema_utils import json_schema_type
 from pydantic import BaseModel, Field, SecretStr
@@ -48,3 +48,10 @@ class NVIDIAConfig(BaseModel):
         default=60,
         description="Timeout for the HTTP requests",
     )
+
+    @classmethod
+    def sample_run_config(cls, **kwargs) -> Dict[str, Any]:
+        return {
+            "url": "https://integrate.api.nvidia.com",
+            "api_key": "${env.NVIDIA_API_KEY}",
+        }

llama_stack/templates/nvidia/__init__.py

@@ -0,0 +1,7 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from .nvidia import get_distribution_template  # noqa: F401

llama_stack/templates/nvidia/build.yaml

@@ -0,0 +1,30 @@
+version: '2'
+name: nvidia
+distribution_spec:
+  description: Use NVIDIA NIM for running LLM inference
+  providers:
+    inference:
+    - remote::nvidia
+    memory:
+    - inline::faiss
+    safety:
+    - inline::llama-guard
+    agents:
+    - inline::meta-reference
+    telemetry:
+    - inline::meta-reference
+    eval:
+    - inline::meta-reference
+    datasetio:
+    - remote::huggingface
+    - inline::localfs
+    scoring:
+    - inline::basic
+    - inline::llm-as-judge
+    - inline::braintrust
+    tool_runtime:
+    - remote::brave-search
+    - remote::tavily-search
+    - inline::code-interpreter
+    - inline::memory-runtime
+image_type: conda

llama_stack/templates/nvidia/doc_template.md

@@ -0,0 +1,61 @@
+# NVIDIA Distribution
+
+The `llamastack/distribution-{{ name }}` distribution consists of the following provider configurations.
+
+{{ providers_table }}
+
+{% if run_config_env_vars %}
+### Environment Variables
+
+The following environment variables can be configured:
+
+{% for var, (default_value, description) in run_config_env_vars.items() %}
+- `{{ var }}`: {{ description }} (default: `{{ default_value }}`)
+{% endfor %}
+{% endif %}
+
+{% if default_models %}
+### Models
+
+The following models are available by default:
+
+{% for model in default_models %}
+- `{{ model.model_id }} ({{ model.provider_model_id }})`
+{% endfor %}
+{% endif %}
+
+
+### Prerequisite: API Keys
+
+Make sure you have access to a NVIDIA API Key. You can get one by visiting [https://build.nvidia.com/](https://build.nvidia.com/).
+
+
+## Running Llama Stack with NVIDIA
+
+You can do this via Conda (build code) or Docker which has a pre-built image.
+
+### Via Docker
+
+This method allows you to get started quickly without having to build the distribution code.
+
+```bash
+LLAMA_STACK_PORT=5001
+docker run \
+  -it \
+  -p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
+  -v ./run.yaml:/root/my-run.yaml \
+  llamastack/distribution-{{ name }} \
+  --yaml-config /root/my-run.yaml \
+  --port $LLAMA_STACK_PORT \
+  --env NVIDIA_API_KEY=$NVIDIA_API_KEY
+```
+
+### Via Conda
+
+```bash
+llama stack build --template nvidia --image-type conda
+llama stack run ./run.yaml \
+  --port 5001 \
+  --env NVIDIA_API_KEY=$NVIDIA_API_KEY
+  --env INFERENCE_MODEL=$INFERENCE_MODEL
+```

llama_stack/templates/nvidia/nvidia.py

@@ -0,0 +1,70 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from pathlib import Path
+
+from llama_stack.distribution.datatypes import ModelInput, Provider
+from llama_stack.providers.remote.inference.nvidia import NVIDIAConfig
+
+from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
+
+
+def get_distribution_template() -> DistributionTemplate:
+    providers = {
+        "inference": ["remote::nvidia"],
+        "memory": ["inline::faiss"],
+        "safety": ["inline::llama-guard"],
+        "agents": ["inline::meta-reference"],
+        "telemetry": ["inline::meta-reference"],
+        "eval": ["inline::meta-reference"],
+        "datasetio": ["remote::huggingface", "inline::localfs"],
+        "scoring": ["inline::basic", "inline::llm-as-judge", "inline::braintrust"],
+        "tool_runtime": [
+            "remote::brave-search",
+            "remote::tavily-search",
+            "inline::code-interpreter",
+            "inline::memory-runtime",
+        ],
+    }
+
+    inference_provider = Provider(
+        provider_id="nvidia",
+        provider_type="remote::nvidia",
+        config=NVIDIAConfig.sample_run_config(),
+    )
+
+    inference_model = ModelInput(
+        model_id="${env.INFERENCE_MODEL}",
+        provider_id="nvidia",
+    )
+
+    return DistributionTemplate(
+        name="nvidia",
+        distro_type="remote_hosted",
+        description="Use NVIDIA NIM for running LLM inference",
+        docker_image=None,
+        template_path=Path(__file__).parent / "doc_template.md",
+        providers=providers,
+        default_models=[inference_model],
+        run_configs={
+            "run.yaml": RunConfigSettings(
+                provider_overrides={
+                    "inference": [inference_provider],
+                },
+                default_models=[inference_model],
+            ),
+        },
+        run_config_env_vars={
+            "LLAMASTACK_PORT": (
+                "5001",
+                "Port for the Llama Stack distribution server",
+            ),
+            "NVIDIA_API_KEY": (
+                "",
+                "NVIDIA API Key",
+            ),
+        },
+    )

llama_stack/templates/nvidia/run.yaml

@@ -0,0 +1,100 @@
+version: '2'
+image_name: nvidia
+conda_env: nvidia
+apis:
+- agents
+- datasetio
+- eval
+- inference
+- memory
+- safety
+- scoring
+- telemetry
+- tool_runtime
+providers:
+  inference:
+  - provider_id: nvidia
+    provider_type: remote::nvidia
+    config:
+      url: https://integrate.api.nvidia.com
+      api_key: ${env.NVIDIA_API_KEY}
+  memory:
+  - provider_id: faiss
+    provider_type: inline::faiss
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/faiss_store.db
+  safety:
+  - provider_id: llama-guard
+    provider_type: inline::llama-guard
+    config: {}
+  agents:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      persistence_store:
+        type: sqlite
+        namespace: null
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/agents_store.db
+  telemetry:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      service_name: ${env.OTEL_SERVICE_NAME:llama-stack}
+      sinks: ${env.TELEMETRY_SINKS:console,sqlite}
+      sqlite_db_path: ${env.SQLITE_DB_PATH:~/.llama/distributions/nvidia/trace_store.db}
+  eval:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config: {}
+  datasetio:
+  - provider_id: huggingface
+    provider_type: remote::huggingface
+    config: {}
+  - provider_id: localfs
+    provider_type: inline::localfs
+    config: {}
+  scoring:
+  - provider_id: basic
+    provider_type: inline::basic
+    config: {}
+  - provider_id: llm-as-judge
+    provider_type: inline::llm-as-judge
+    config: {}
+  - provider_id: braintrust
+    provider_type: inline::braintrust
+    config:
+      openai_api_key: ${env.OPENAI_API_KEY:}
+  tool_runtime:
+  - provider_id: brave-search
+    provider_type: remote::brave-search
+    config:
+      api_key: ${env.BRAVE_SEARCH_API_KEY:}
+      max_results: 3
+  - provider_id: tavily-search
+    provider_type: remote::tavily-search
+    config:
+      api_key: ${env.TAVILY_SEARCH_API_KEY:}
+      max_results: 3
+  - provider_id: code-interpreter
+    provider_type: inline::code-interpreter
+    config: {}
+  - provider_id: memory-runtime
+    provider_type: inline::memory-runtime
+    config: {}
+metadata_store:
+  type: sqlite
+  db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/registry.db
+models:
+- metadata: {}
+  model_id: ${env.INFERENCE_MODEL}
+  provider_id: nvidia
+  model_type: llm
+shields: []
+memory_banks: []
+datasets: []
+scoring_fns: []
+eval_tasks: []
+tool_groups: []