From 610dea84c20b59bb4a02d6574055ed073d12e916 Mon Sep 17 00:00:00 2001 From: Matthew Farrellee Date: Fri, 21 Feb 2025 15:41:49 -0600 Subject: [PATCH] add nvidia embedding implementation for new signature, task_type, output_dimention, text_truncation --- .../remote/inference/nvidia/nvidia.py | 32 ++++-- tests/client-sdk/inference/test_embedding.py | 100 +++++++++++++++++- 2 files changed, 124 insertions(+), 8 deletions(-) diff --git a/llama_stack/providers/remote/inference/nvidia/nvidia.py b/llama_stack/providers/remote/inference/nvidia/nvidia.py index ecd53e91c..4c731a365 100644 --- a/llama_stack/providers/remote/inference/nvidia/nvidia.py +++ b/llama_stack/providers/remote/inference/nvidia/nvidia.py @@ -142,18 +142,38 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper): # # we can ignore str and always pass List[str] to OpenAI # - flat_contents = [ - item.text if isinstance(item, TextContentItem) else item - for content in contents - for item in (content if isinstance(content, list) else [content]) - ] + 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) + extra_body = {} + + if text_truncation is not None: + text_truncation_options = { + TextTruncation.none: "NONE", + TextTruncation.end: "END", + TextTruncation.start: "START", + } + if text_truncation not in text_truncation_options: + raise ValueError(f"Invalid text_truncation: {text_truncation}") + extra_body["truncate"] = text_truncation_options[text_truncation] + + if output_dimension is not None: + extra_body["dimensions"] = output_dimension + + if task_type is not None: + task_type_options = { + EmbeddingTaskType.document: "DOCUMENT", + EmbeddingTaskType.query: "QUERY", + } + if task_type not in task_type_options: + raise ValueError(f"Invalid task_type: {task_type}") + extra_body["input_type"] = task_type_options[task_type] + response = await self._client.embeddings.create( model=model, input=input, - # extra_body={"input_type": "passage"|"query"}, # TODO(mf): how to tell caller's intent? + extra_body=extra_body, ) # diff --git a/tests/client-sdk/inference/test_embedding.py b/tests/client-sdk/inference/test_embedding.py index 3304406a9..2d168b525 100644 --- a/tests/client-sdk/inference/test_embedding.py +++ b/tests/client-sdk/inference/test_embedding.py @@ -14,6 +14,23 @@ # - array of a text (TextContentItem) # Types of output: # - list of list of floats +# Params: +# - text_truncation +# - absent w/ long text -> error +# - none w/ long text -> error +# - absent w/ short text -> ok +# - none w/ short text -> ok +# - end w/ long text -> ok +# - end w/ short text -> ok +# - start w/ long text -> ok +# - start w/ short text -> ok +# - output_dimension +# - response dimension matches +# - task_type, only for asymmetric models +# - query embedding != passage embedding +# Negative: +# - long string +# - long text # # Todo: # - negative tests @@ -23,8 +40,6 @@ # - empty text # - empty image # - long -# - long string -# - long text # - large image # - appropriate combinations # - batch size @@ -48,10 +63,14 @@ from llama_stack_client.types.shared.interleaved_content import ( TextContentItem, ) +from llama_stack.apis.inference import EmbeddingTaskType, TextTruncation + DUMMY_STRING = "hello" DUMMY_STRING2 = "world" +DUMMY_LONG_STRING = "NVDA " * 10240 DUMMY_TEXT = TextContentItem(text=DUMMY_STRING, type="text") DUMMY_TEXT2 = TextContentItem(text=DUMMY_STRING2, type="text") +DUMMY_LONG_TEXT = TextContentItem(text=DUMMY_LONG_STRING, type="text") # TODO(mf): add a real image URL and base64 string DUMMY_IMAGE_URL = ImageContentItem( image=ImageContentItemImage(url=ImageContentItemImageURL(uri="https://example.com/image.jpg")), type="image" @@ -96,3 +115,80 @@ def test_embedding_image(llama_stack_client, embedding_model_id, contents): assert len(response.embeddings) == sum(len(content) if isinstance(content, list) else 1 for content in contents) assert isinstance(response.embeddings[0], list) assert isinstance(response.embeddings[0][0], float) + + +@pytest.mark.parametrize( + "text_truncation", + [ + TextTruncation.end, + TextTruncation.start, + ], +) +@pytest.mark.parametrize( + "contents", + [ + [DUMMY_LONG_TEXT], + [DUMMY_STRING], + ], + ids=[ + "long", + "short", + ], +) +def test_embedding_truncation(llama_stack_client, embedding_model_id, text_truncation, contents): + response = llama_stack_client.inference.embeddings( + model_id=embedding_model_id, contents=contents, text_truncation=text_truncation + ) + assert isinstance(response, EmbeddingsResponse) + assert len(response.embeddings) == 1 + assert isinstance(response.embeddings[0], list) + assert isinstance(response.embeddings[0][0], float) + + +@pytest.mark.parametrize( + "text_truncation", + [ + None, + TextTruncation.none, + ], +) +@pytest.mark.parametrize( + "contents", + [ + [DUMMY_LONG_TEXT], + [DUMMY_LONG_STRING], + ], + ids=[ + "long-text", + "long-str", + ], +) +def test_embedding_truncation_error(llama_stack_client, embedding_model_id, text_truncation, contents): + response = llama_stack_client.inference.embeddings( + model_id=embedding_model_id, contents=[DUMMY_LONG_TEXT], text_truncation=text_truncation + ) + assert isinstance(response, EmbeddingsResponse) + assert len(response.embeddings) == 1 + assert isinstance(response.embeddings[0], list) + assert isinstance(response.embeddings[0][0], float) + + +@pytest.mark.xfail(reason="Only valid for model supporting dimension reduction") +def test_embedding_output_dimension(llama_stack_client, embedding_model_id): + base_response = llama_stack_client.inference.embeddings(model_id=embedding_model_id, contents=[DUMMY_STRING]) + test_response = llama_stack_client.inference.embeddings( + model_id=embedding_model_id, contents=[DUMMY_STRING], output_dimension=32 + ) + assert len(base_response.embeddings[0]) != len(test_response.embeddings[0]) + assert len(test_response.embeddings[0]) == 32 + + +@pytest.mark.xfail(reason="Only valid for model supporting task type") +def test_embedding_task_type(llama_stack_client, embedding_model_id): + query_embedding = llama_stack_client.inference.embeddings( + model_id=embedding_model_id, contents=[DUMMY_STRING], task_type=EmbeddingTaskType.query + ) + document_embedding = llama_stack_client.inference.embeddings( + model_id=embedding_model_id, contents=[DUMMY_STRING], task_type=EmbeddingTaskType.document + ) + assert query_embedding.embeddings != document_embedding.embeddings