feat: NVIDIA allow non-llama model registration

[raspawar]

What does this PR do?

Adds custom model registration functionality to NVIDIAInferenceAdapter which let's the inference happen on:

• post-training model
• non-llama models in API Catalogue(behind https://integrate.api.nvidia.com and endpoints compatible with AyncOpenAI)

Example Usage:

from llama_stack.apis.models import Model, ModelType
from llama_stack.distribution.library_client import LlamaStackAsLibraryClient
client = LlamaStackAsLibraryClient("nvidia")
_ = client.initialize()

client.models.register(
        model_id=model_name,
        model_type=ModelType.llm,
        provider_id="nvidia"
)

response = client.inference.chat_completion(
    model_id=model_name,
    messages=[{"role":"system","content":"You are a helpful assistant."},{"role":"user","content":"Write a limerick about the wonders of GPU computing."}],
)

Test Plan

pytest tests/unit/providers/nvidia/test_supervised_fine_tuning.py 
========================================================== test session starts ===========================================================
platform linux -- Python 3.10.0, pytest-8.3.5, pluggy-1.5.0
rootdir: /home/ubuntu/llama-stack
configfile: pyproject.toml
plugins: anyio-4.9.0
collected 6 items                                                                                                                        

tests/unit/providers/nvidia/test_supervised_fine_tuning.py ......                                                                  [100%]

============================================================ warnings summary ============================================================
../miniconda/envs/nvidia-1/lib/python3.10/site-packages/pydantic/fields.py:1076
  /home/ubuntu/miniconda/envs/nvidia-1/lib/python3.10/site-packages/pydantic/fields.py:1076: PydanticDeprecatedSince20: Using extra keyword arguments on `Field` is deprecated and will be removed. Use `json_schema_extra` instead. (Extra keys: 'contentEncoding'). Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
    warn(

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
====================================================== 6 passed, 1 warning in 1.51s ======================================================

Updated Readme.md

cc: @dglogo, @sumitb, @mattf

[mattf]

thanks for adding this. few comments inline for you.

[mattf]

this is a behavior change.

current behavior: always add /v1
new behavior: add /v1 for hosted and don't for non-hosted

the behavior should be consistent between hosted and non-hosted. for instance, a user should not need to know they're talking to https://integrate.api.nv.c and therefore don't need to supply the /v1 or since they're talking to http://localhost that they do need to provide the /v1.

is there an issue w/ /v1 and customizer?

[raspawar]

Adding /v1 may produce errors, specially when user is specifying NVIDIA_BASE_URL
Can we remove the /v1 entirely? And add it in default base_url, what do we miss in that case?
As some models endpoints on API catalogue follow /chat/completion.

[mattf]

this should be handled within the provider model id function

[mattf]

i believe model is a https://github.com/meta-llama/llama-stack/blob/main/llama_stack/apis/models/models.py#L31, which does not have a metadata

https://github.com/meta-llama/llama-stack/blob/main/llama_stack/models/llama/datatypes.py#L346 has a metadata, and confusingly the same class name.

suggestion: trust the input config. it should fail at inference if it's incorrect.

[raspawar]

This is base register_model() logic: https://github.com/meta-llama/llama-stack/blob/main/llama_stack/providers/utils/inference/model_registry.py#L76

I modified only parts to allow non-llama models.