2

I have created the following piece of code using Jupyter Notebook and langchain==0.0.134 (which in my case comes with openai==0.27.2). The code takes a CSV file and loads it in Chroma using OpenAI Embeddings.

CSV

COLUMN1;COLUMN2
Hello;World
From;CSV

Jupyter Notebook

#!/usr/bin/env python
# coding: utf-8
get_ipython().run_line_magic('load_ext', 'dotenv')
get_ipython().run_line_magic('dotenv', '')

# ### CSV Load
from langchain.document_loaders.csv_loader import CSVLoader

csv_args = {"delimiter": ";",
            "quotechar": '"',
           'fieldnames': ['COLUMN1','COLUMN2']}
loader = CSVLoader(file_path='./data/stack-overflow-test.csv', csv_args=csv_args)

# ### Load in Chroma
from langchain.vectorstores import Chroma
from langchain.indexes import VectorstoreIndexCreator
from langchain.embeddings.openai import OpenAIEmbeddings

index_creator = VectorstoreIndexCreator(
    vectorstore_cls=Chroma,
    embedding=OpenAIEmbeddings(),
    vectorstore_kwargs= {"collection_name": "collection"}
)

# This is the line of code that is recorded with the "packet analyzer"
indexWrapper = index_creator.from_loaders([loader])

If I check the request (using Wireshark), I obtain the following:

Request

POST /v1/engines/text-embedding-ada-002/embeddings HTTP/1.1
Host: api.openai.com
User-Agent: OpenAI/v1 PythonBindings/0.27.2
Content-Type: application/json

{
  "input": [
    [82290, 16, 25, 76880, 82290, 16, 40123, 17, 25, 40123, 17],
    [82290, 16, 25, 22691, 40123, 17, 25, 4435],
    [82290, 16, 25, 5659, 40123, 17, 25, 28545]
  ],
  "encoding_format": "base64"
}

Reply

openai-version: 2020-10-01
Content-Type: application/json
{
    "object": "list",
    "data": [
      {
        "object": "embedding",
        "index": 0,
        "embedding": "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"
      },
      {
        "object": "embedding",
        "index": 1,
        "embedding": "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"
      },
      {
        "object": "embedding",
        "index": 2,
        "embedding": "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"
      }
    ],
    "model": "text-embedding-ada-002-v2",
    "usage": {
      "prompt_tokens": 27,
      "total_tokens": 27
    }
  }

I want to understand what is happening behind the scenes as this kind of debugging is useful for troubleshooting. As you can see, the payload has an input field with a matrix of numbers, but it does not make sense to me (it does not match the documentation).

So I have two questions:

  1. Why does the input field have this matrix of numbers?
  2. How can I decode the answer? I couldn't create the vector I am supposed to receive when I decode the embedding field from the answer using Base64.

It looks like the Python client from OpenAI uses an older version of the API (can be that the reason? I didn't use the API before).

ChatGPT mentioned

The tokens are represented by numerical IDs such as 82290, 16, 25, etc., which likely correspond to a vocabulary or tokenization scheme used by OpenAI

However, it does not provide references and I would like to have them. It might be related to one of this tools Tiktoken, Huggingface Tokenizer

Edu
  • 159
  • 1
  • 14

0 Answers0