Hi! does anybody knows where I can find info about the different types of queries (Synthesis, Routing, Compare/Contrast, etc) and what are the best use cases for each of them?

Hi, I'm having an error from version 0.5.9

GPTSimpleVectorIndex.load_from_disk('path_to_my_index')
is failing in /llama_index/vector_stores/registry.py with KeyError: '__type__'

I was checking the changelog of 0.5.9 but I can't figure what's the error, I gues it's is related with the change in gpt_index/constants.py :
VECTOR_STORE_CONFIG_DICT_KEY = "vector_store" was removed and these two keys were added:


but that's all I can see

hey, does anyone experienced "wrong" answers from index queries? After I created the index with a .md document, I ask a very specific question and the answer comes from articles in the internet and not from my document. I'm asking who is the author of the document and the document has a section literally saying "The authors are..." but I'm getting wrong authors from other sources related with the topic. I tried making the query with mode="embedding" but still not getting right answers for specific questions. I'm pretty sure I am missing something here. Thanks

Hi guys, I'm new here. First of all congrats for such a great project.
I'm currently trying to create an index using GPTSimpleVectorIndex and I have two issues at the moment:

1. When passing model_name parameter with text-davinci-003, in my openai usage page I always see that text-embedding-ada-002-v2 was used. I'm currently in the Free trial usage, not sure if that's related or if davinci is used for the queries only.
2. I wanted to analyse tokens usage with no costs before creating the whole index, so I was using the MockLLMPredictor with GPTSimpleVectorIndex but is still using tokens from openai even though llm_predictor.last_token_usage reports 0. I'm not sure if I'll be charged for create an index or not and with which model.

Thanks 🙂