search

brianpetro / obsidian-smart-connections

Chat with your notes & see links to related content with AI embeddings. Use local models or 100+ via APIs like Claude, Gemini, ChatGPT & Llama 3
https://smartconnections.app
GNU General Public License v3.0
2.79k stars 183 forks source link

Being told block not found #872

Open bbecausereasonss opened 6 days ago

bbecausereasonss commented 6 days ago

Gemini Latest, giving me this when I ask it to compare 2 notes:

Based on your notes, I am still unable to provide a comparison of the two webinars. All linked notes return "BLOCK NOT FOUND," indicating there's no content to analyze. I need the actual content of your notes about the webinars to understand their differences. Simply providing links to non-existent files doesn't give me the information required to answer your question.

Mind you both notes have embeddings and pull-up (similarity tree)

image

Used to be able to choose a different model to embed blocks; but all I see is an option (embed blocks using embedding model) but if I un click it, I don't see a place to choose which model to use?

bbecausereasonss commented 6 days ago

Same error in Claude now.

image

bbecausereasonss commented 6 days ago

Just upgraded to 2.2 (same issue) very odd.

bbecausereasonss commented 5 days ago

I completely re did all the embeddings; now getting this... very frustrating. Might go back to super old version ;/

image

bbecausereasonss commented 5 days ago

Reverted to stable release... I all of a sudden cannot do the simplest things in chat which I was able to do fine before; not sure what changed... Even in OpenAI now... this is just after a second prompt based on a webinar transcript :///

Clearly something wrong with the embeddings??

API Error: This model's maximum context length is 128000 tokens. However, your messages resulted in 133415 tokens (133203 in the messages, 212 in the functions). Please reduce the length of the messages or functions.

bbecausereasonss commented 4 days ago

Again, worked for 2-3 generations now...

image

Update. After this error, I retried the same prompt and it went through... so odd. (3.5 Sonnet). on 2.2.8.0

brianpetro commented 4 days ago

The rate limit issue is known and will be addressed in the successor to the current Smart Chat which is a WIP, and will likely be one of the first things I address. This was supposed to happen this week but... (I'm not much into excuses, so imagine a sufficient one goes here).

The second one looks like it's probably just an Anthropic API failure (which of course should be better handled in the future). And the block not found warnings just prior to the error, while interesting, should not have an impact (indicated by yellow highlight).

The "Prune" button in the settings should handle those missing blocks by deleting them since there isn't a valid source document/content (indicated by the error)

Thanks for the helpful screenshots 🌴