local ai spam analysis with ollama + litellm + python + notmuch + prefect
Discussion
going to see if local ai can efficiently manage my inbox. Thanks for the idea nostr:npub13ql75nq8rldygpkjke47y893akh5tglqtqzs6cspancaxktthsusvfqcg7
Good idea
What local ai can do ? Can someone teach me
email clients like gmail have been trying their best at doing this perpetually
i think use ollama , example prompt : "classify this email as 'spam' or 'ham'. return only the label . Email :{cleaned_text}.
querry ollame with phyton use your api ..is that how it works ?
it's actually very good idea , everything run on your local computer , using ollama , no data leak, no cloud .
simple pipeline, no need to use ML pipeline , the LLM do the heavy lfiting
this probably solve my problem , when gmail said : gmail has scanned your email , i felt a bit stunned ...gmail know i just making purchased sats on primal 😂
please tell me hoe it goes! I downloaded the Llama open source thing and must admit: out of the box it is very crappy.
What does the timeline flowchart show? Is it a screenshot of one of the tools you've mentioned?
yeah that's prefect. you decorate python functions and then it adds a workflow graph with generated artifacts
this is my entire script that is currently working: