Hey. I'm getting this weird key error

How did you build your index?

https://github.com/run-llama/llama_index/issues/1769

i referred to this, but i made the index in a fresh notebook

Seems to work for me

https://colab.research.google.com/drive/1vmeIUYmlUHodixrsYYzviHAUcaoLTTvl?usp=sharing

It worked for me with a pdf. I was thinking it was a issue with html's. I tried another file, but i got the same error. Currently trying on another html file in a new notebook, it takes time to load llama

I'll try with a few other files

Hmm tried with a random html file as well, still worked

If I could reproduce, I could look into it further

alright. I will try a file or two more, if I get it again, share you the notebook in your DM if you don't mind?

If this doesn't work, I'm planning to get text from html directly(beautifulsoup), and then find a way to create embeddings and store it in my index.

Yea feel free to share any notebooks that reproduce the issue 🙏 Would love to know why that happens.

The error to me somehow indicates that the vector store and docstore do not contain the same data (somehow)

Hey. Tried all of it in a fresh notebook, turns out it gives the error only with that specfic file. Wasnt expecting this though

A different html file worked

made index with both the files back and forth in the same session, to confirm

Hey. So does just defining my openapi key in env will make the query engine use gpt 3.5? (because i dont see any llm defined here)

yea 3.5 is the default -- and text-embedding-ada-002 is the default embedding model

understood!

Just so that anyone stumbles here for help, I used the same document but in chroma db and didnt get any errors