structured output for /completion API ollama

llama_stack/providers/remote/inference/ollama/ollama.py

@@ -172,6 +172,7 @@ class OllamaInferenceAdapter(Inference, ModelsProtocolPrivate):
             model=model.provider_resource_id,
             content=content,
             sampling_params=sampling_params,
+            response_format=response_format,
             stream=stream,
             logprobs=logprobs,
         )

llama_stack/providers/tests/inference/test_text_inference.py

@@ -208,7 +208,6 @@ class TestInference:
                 assert not chunk.logprobs, "Logprobs should be empty"
 
     @pytest.mark.asyncio(loop_scope="session")
-    @pytest.mark.skip("This test is not quite robust")
     async def test_completion_structured_output(self, inference_model, inference_stack):
         inference_impl, _ = inference_stack
 

tests/client-sdk/inference/test_inference.py

@@ -11,7 +11,7 @@ import pytest
 from pydantic import BaseModel
 
 PROVIDER_TOOL_PROMPT_FORMAT = {
-    "remote::ollama": "python_list",
+    "remote::ollama": "json",
     "remote::together": "json",
     "remote::fireworks": "json",
 }