AttributeError: 'OpenAIBackend' object has no attribute 'encode'

Nobody has claimed this yet.

Hi,
I still get this error message despite I use the last commit & version from bertopic, could you please help ?

Code:

import openai
from bertopic.backend import OpenAIBackend
from openai import AzureOpenAI
client = AzureOpenAI(
api_version = "2023-09-15-preview",
api_key =os.getenv("OPENAI_API_KEY").strip(),
azure_endpoint ="https://x-.openai.azure.com/",
)
embedding_model = OpenAIBackend(client,"text-embedding-3-large")

embeddings = embedding_model.encode(df['PreprocessedText'].tolist(), show_progress_bar=True)

from bertopic import BERTopic
# Initialize and train BERTopic model
topic_model = BERTopic(
embedding_model=embedding_model,
vectorizer_model=vectorizer_model,
umap_model=umap_model,
calculate_probabilities=True,
#hdbscan_model=hdbscan_model,
representation_model=representation_model,
verbose=True,
nr_topics=10
)

# Fit the topic model and transform the data
topics, probs = topic_model.fit_transform(df['PreprocessedText'].values)

Start by checking the BERTopic OpenAIBackend API and the installed BERTopic version against the reported encode call, then reproduce the example with AzureOpenAI. The issue is done when the embedding backend can be used with the shown workflow and BERTopic fit_transform completes without the AttributeError.

Written by the indexing model from the issue text.