Local LLM with LM Studio Server: Error in POST payload when using langchain_openai.OpenAIEmbeddings for embedding API.

[relusion]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

# this code works just fine with local deployed embedding model with LM Studio server with OPENAI API.
from openai import OpenAI

client = OpenAI(base_url="http://localhost:8999/v1", api_key="lm-studio")

def get_embeddings(texts, model="nomic-ai/nomic-embed-text-v1.5-GGUF"):
    texts = [text.replace("\n", " ") for text in texts]
    return client.embeddings.create(input=texts, model=model).data

print(get_embeddings(["how to find out how LLM applications are performing in real-world scenarios?"]))

This is what I see on the server side and the server returns embedding data back to the code:

[2024-05-06 13:51:34.227] [INFO] Received POST request to /v1/embeddings with body:
{
  "input": [
    "how to find out how LLM applications are performing in real-world scenarios?"
  ],
  "model": "nomic-ai/nomic-embed-text-v1.5-GGUF",
  "encoding_format": "base64"
}

#however if I switch to OpenAIEmbeddings, this code does not work
from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings(openai_api_key="sk-1234", base_url="http://localhost:8999/v1", model="nomic-ai/nomic-embed-text-v1.5-GGUF")
test = embeddings.embed_query("how to find out how LLM applications are performing in real-world scenarios?")

# this is what I see on the server side:
[2024-05-06 13:52:08.629] [INFO] Received POST request to /v1/embeddings with body:
{
  "input": [
    [
      5269,
      311,
      1505,
      704,
      1268,
      445,
      11237,
      8522,
      527,
      16785,
      304,
      1972,
      31184,
      26350,
      30
    ]
  ],
  "model": "nomic-ai/nomic-embed-text-v1.5-GGUF",
  "encoding_format": "base64"
}

Error Message and Stack Trace (if applicable)

Error on the server side:
[ERROR] 'input' field must be a string or an array of strings

Description

I encountered an issue with the langchain_openai library where using OpenAIEmbeddings to embed a text query results in a malformed POST request payload to the API endpoint. Below is a comparison of the expected and actual requests.

System Info

System Information

OS: Windows
OS Version: 10.0.22631
Python Version: 3.12.3 (tags/v3.12.3:f6650f9, Apr 9 2024, 14:05:25) [MSC v.1938 64 bit (AMD64)]

Package Information

langchain_core: 0.1.50
langchain: 0.1.17
langchain_community: 0.0.36
langsmith: 0.1.54
langchain_chroma: 0.1.0
langchain_openai: 0.1.6
langchain_text_splitters: 0.0.1

Packages not installed (Not Necessarily a Problem)

The following packages were not found:

langgraph
langserve

[relusion]

An observation:
Setting check_embedding_ctx_length to False fixes the problem.

OpenAIEmbeddings(check_embedding_ctx_length=False, openai_api_key="sk-1234", base_url="http://localhost:8999/v1",model="nomic-ai/nomic-embed-text-v1.5-GGUF")

[vickybiswas]

I have the same issue, and it is not recognizing the suggested parameter
Embeddings.create() got an unexpected keyword argument 'check_embedding_ctx_length'

[ticoAg]

I have the same issue, and it is not recognizing the suggested parameter Embeddings.create() got an unexpected keyword argument 'check_embedding_ctx_length'

upgrade package and check usage

from langchain_openai import OpenAIEmbeddings
model = OpenAIEmbeddings(model="Your embed model", check_embedding_ctx_length=False)
response = model.embed_query("Hello world")
print(response)

[Trinkes]

hey @ticoAg, thanks for the suggestion!
The check_embedding_ctx_length=False worked for me but is this going to be fixed? It sounds like a compatibility issue.