This commit is contained in:
Timothy J. Baek 2024-10-25 22:23:21 -07:00
parent 925a903e38
commit f2c78ac0fb

View File

@ -667,7 +667,11 @@ def save_docs_to_vector_db(
add_start_index=True, add_start_index=True,
) )
elif app.state.config.TEXT_SPLITTER == "token": elif app.state.config.TEXT_SPLITTER == "token":
tiktoken.get_encoding(app.state.config.TIKTOKEN_ENCODING_NAME) log.info(
f"Using token text splitter: {app.state.config.TIKTOKEN_ENCODING_NAME}"
)
tiktoken.get_encoding(str(app.state.config.TIKTOKEN_ENCODING_NAME))
text_splitter = TokenTextSplitter( text_splitter = TokenTextSplitter(
encoding_name=str(app.state.config.TIKTOKEN_ENCODING_NAME), encoding_name=str(app.state.config.TIKTOKEN_ENCODING_NAME),
chunk_size=app.state.config.CHUNK_SIZE, chunk_size=app.state.config.CHUNK_SIZE,