Find answers from the community

Updated 3 months ago

Graph index

JJack2020

my code looks liek this: for text in texts:
if self.find_matching_keyword(text, key_words_for_remove_unneeded_text):
dc = GPTListIndex.from_documents([Document(text)])
tree_index_list.append(dc)
tree_index_summary.append(str(text))

loop_time = time.time()
print("Loop time: {:.6f} seconds".format(loop_time - filter_time))

4 comments

LLogan M

So your graph is a bunch of list indexes with a tree index on top?

JJack2020

yes...

JJack2020

does it help to chunk them into smaller ones?

LLogan M

So, with a list index, it will check EVERY node in that list index. So if there are a lot of documents in there, it will take some time

Typically a list index is best for generating summaries. Maybe you'd have more luck using vector indexes instead?

Add a reply