mirror of
https://github.com/hwchase17/langchain.git
synced 2025-08-18 09:01:03 +00:00
56499cf58b
From what I can tell response using SDK is not deterministic:
```python
import numpy as np
import openai
documents = ["disallowed special token '<|endoftext|>'"]
model = "text-embedding-ada-002"
direct_output_1 = (
openai.OpenAI()
.embeddings.create(input=documents, model=model)
.data[0]
.embedding
)
for i in range(10):
direct_output_2 = (
openai.OpenAI()
.embeddings.create(input=documents, model=model)
.data[0]
.embedding
)
print(f"{i}: {np.isclose(direct_output_1, direct_output_2).all()}")
```
```
0: True
1: True
2: True
3: True
4: False
5: True
6: True
7: True
8: True
9: True
```
See related discussion here:
https://community.openai.com/t/can-text-embedding-ada-002-be-made-deterministic/318054
Found the same result using `"text-embedding-3-small"`.
70 lines
2.1 KiB
Python
70 lines
2.1 KiB
Python
"""Test OpenAI embeddings."""
|
|
|
|
import numpy as np
|
|
import openai
|
|
|
|
from langchain_openai.embeddings.base import OpenAIEmbeddings
|
|
|
|
|
|
def test_langchain_openai_embedding_documents() -> None:
|
|
"""Test openai embeddings."""
|
|
documents = ["foo bar"]
|
|
embedding = OpenAIEmbeddings()
|
|
output = embedding.embed_documents(documents)
|
|
assert len(output) == 1
|
|
assert len(output[0]) > 0
|
|
|
|
|
|
def test_langchain_openai_embedding_query() -> None:
|
|
"""Test openai embeddings."""
|
|
document = "foo bar"
|
|
embedding = OpenAIEmbeddings()
|
|
output = embedding.embed_query(document)
|
|
assert len(output) > 0
|
|
|
|
|
|
def test_langchain_openai_embeddings_dimensions() -> None:
|
|
"""Test openai embeddings."""
|
|
documents = ["foo bar"]
|
|
embedding = OpenAIEmbeddings(model="text-embedding-3-small", dimensions=128)
|
|
output = embedding.embed_documents(documents)
|
|
assert len(output) == 1
|
|
assert len(output[0]) == 128
|
|
|
|
|
|
def test_langchain_openai_embeddings_equivalent_to_raw() -> None:
|
|
documents = ["disallowed special token '<|endoftext|>'"]
|
|
embedding = OpenAIEmbeddings()
|
|
|
|
lc_output = embedding.embed_documents(documents)[0]
|
|
direct_output = (
|
|
openai.OpenAI()
|
|
.embeddings.create(input=documents, model=embedding.model)
|
|
.data[0]
|
|
.embedding
|
|
)
|
|
assert np.allclose(lc_output, direct_output, atol=0.001)
|
|
|
|
|
|
async def test_langchain_openai_embeddings_equivalent_to_raw_async() -> None:
|
|
documents = ["disallowed special token '<|endoftext|>'"]
|
|
embedding = OpenAIEmbeddings()
|
|
|
|
lc_output = (await embedding.aembed_documents(documents))[0]
|
|
client = openai.AsyncOpenAI()
|
|
direct_output = (
|
|
(await client.embeddings.create(input=documents, model=embedding.model))
|
|
.data[0]
|
|
.embedding
|
|
)
|
|
assert np.allclose(lc_output, direct_output, atol=0.001)
|
|
|
|
|
|
def test_langchain_openai_embeddings_dimensions_large_num() -> None:
|
|
"""Test openai embeddings."""
|
|
documents = [f"foo bar {i}" for i in range(2000)]
|
|
embedding = OpenAIEmbeddings(model="text-embedding-3-small", dimensions=128)
|
|
output = embedding.embed_documents(documents)
|
|
assert len(output) == 2000
|
|
assert len(output[0]) == 128
|