2024-02-20 19:22:27 +00:00
|
|
|
|
"""Test SparkLLM Text Embedding."""
|
|
|
|
|
from langchain_community.embeddings.sparkllm import SparkLLMTextEmbeddings
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_baichuan_embedding_documents() -> None:
|
|
|
|
|
"""Test SparkLLM Text Embedding for documents."""
|
|
|
|
|
documents = [
|
|
|
|
|
"iFLYTEK is a well-known intelligent speech and artificial intelligence "
|
|
|
|
|
"publicly listed company in the Asia-Pacific Region. Since its establishment,"
|
|
|
|
|
"the company is devoted to cornerstone technological research "
|
|
|
|
|
"in speech and languages, natural language understanding, machine learning,"
|
|
|
|
|
"machine reasoning, adaptive learning, "
|
|
|
|
|
"and has maintained the world-leading position in those "
|
|
|
|
|
"domains. The company actively promotes the development of A.I. "
|
|
|
|
|
"products and their sector-based "
|
|
|
|
|
"applications, with visions of enabling machines to listen and speak, "
|
|
|
|
|
"understand and think, "
|
|
|
|
|
"creating a better world with artificial intelligence."
|
|
|
|
|
]
|
2024-05-13 18:55:07 +00:00
|
|
|
|
embedding = SparkLLMTextEmbeddings() # type: ignore[call-arg]
|
2024-02-20 19:22:27 +00:00
|
|
|
|
output = embedding.embed_documents(documents)
|
|
|
|
|
assert len(output) == 1 # type: ignore[arg-type]
|
|
|
|
|
assert len(output[0]) == 2560 # type: ignore[index]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_baichuan_embedding_query() -> None:
|
|
|
|
|
"""Test SparkLLM Text Embedding for query."""
|
|
|
|
|
document = (
|
|
|
|
|
"iFLYTEK Open Platform was launched in 2010 by iFLYTEK as China’s "
|
|
|
|
|
"first Artificial Intelligence open platform for Mobile Internet "
|
|
|
|
|
"and intelligent hardware developers"
|
|
|
|
|
)
|
2024-05-13 18:55:07 +00:00
|
|
|
|
embedding = SparkLLMTextEmbeddings() # type: ignore[call-arg]
|
2024-02-20 19:22:27 +00:00
|
|
|
|
output = embedding.embed_query(document)
|
|
|
|
|
assert len(output) == 2560 # type: ignore[arg-type]
|