sentence-transformers How to use ondayex/jina-embed-base-dense-retriever with sentence-transformers:
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("ondayex/jina-embed-base-dense-retriever")
sentences = [
"Best practices for async test_similarity_search_with_relevance_score_with_threshold_and_filter",
"def test_tool_retry_custom_failure_formatter() -> None:\n \"\"\"Test ToolRetryMiddlewarewith custom failure message formatter.\"\"\"\n\n def custom_formatter(exc: Exception) -> str:\n return f\"Custom error: {type(exc).__name__}\"\n\n model = FakeToolCallingModel(\n tool_calls=[\n [ToolCall(name=\"failing_tool\", args={\"value\": \"test\"}, id=\"1\")],\n [],\n ]\n )\n\n retry = ToolRetryMiddleware(\n max_retries=1,\n initial_delay=0.01,\n jitter=False,\n on_failure=custom_formatter,\n )\n\n agent = create_agent(\n model=model,\n tools=[failing_tool],\n middleware=[retry],\n checkpointer=InMemorySaver(),\n )\n\n result = agent.invoke(\n {\"messages\": [HumanMessage(\"Use failing tool\")]},\n {\"configurable\": {\"thread_id\": \"test\"}},\n )\n\n tool_messages = [m for m in result[\"messages\"] if isinstance(m, ToolMessage)]\n assert len(tool_messages) == 1\n assert \"Custom error: ValueError\" in tool_messages[0].content",
"def test_parse_scores(answer: str) -> None:\n result = output_parser.parse(answer)\n\n assert result[\"answer\"] == \"foo bar answer.\"\n\n score = int(result[\"score\"])\n assert score == 80",
"async def test_similarity_search_with_relevance_score_with_threshold_and_filter(\n vector_name: str | None,\n qdrant_location: str,\n) -> None:\n \"\"\"Test end to end construction and search.\"\"\"\n texts = [\"foo\", \"bar\", \"baz\"]\n metadatas = [\n {\"page\": i, \"metadata\": {\"page\": i + 1, \"pages\": [i + 2, -1]}}\n for i in range(len(texts))\n ]\n docsearch = Qdrant.from_texts(\n texts,\n ConsistentFakeEmbeddings(),\n metadatas=metadatas,\n vector_name=vector_name,\n location=qdrant_location,\n )\n score_threshold = 0.99 # for almost exact match\n # test negative filter condition\n negative_filter = {\"page\": 1, \"metadata\": {\"page\": 2, \"pages\": [3]}}\n kwargs = {\"filter\": negative_filter, \"score_threshold\": score_threshold}\n output = docsearch.similarity_search_with_relevance_scores(\"foo\", k=3, **kwargs)\n assert len(output) == 0\n # test positive filter condition\n positive_filter = {\"page\": 0, \"metadata\": {\"page\": 1, \"pages\": [2]}}\n kwargs = {\"filter\": positive_filter, \"score_threshold\": score_threshold}\n output = await docsearch.asimilarity_search_with_relevance_scores(\n \"foo\", k=3, **kwargs\n )\n assert len(output) == 1\n assert all(score >= score_threshold for _, score in output)"
]
embeddings = model.encode(sentences)
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [4, 4] 
