Okay we have downloading, transcribing, and advertisement tagging and summaries working
Idk how useful summaries are they're kind of too high level
Maybe useful for the vector db to narrow down what episodes to query.
Idk wouldn't the vector db take care of that anyway?
This is kind of a combo if two goals:
- Process podcasts so I never hear an ad again
- turn entire series into vector db for RAG
Like "chat with all of RHR"
John
npub1hjle...zg68
ML Engineer
(Former) attorney
Wouldn't it be cool if you could create a knowledge base from an entire podcast series
Download all the episodes, transcribe, extract key insights, vector db, rag Q&A
How can I manifest myself out of these allergies
Kefir marinaded chicken ๐
I can finally run a sonnet 3.5 level model locally ๐
At 6 t/a ๐ซ
Kinda want to make a solar distill and oven now
The latest Huberman-lab guest said essentially that hunting dogs like to hunt BC theyre predators, so
Do prey animals enjoy being chased?
Nourishing neural networks
Knowledge base that's a collection of all your debugging issues and solutions
Just need to bundle up your chat history nicely and share it as a vector db
Like why bother making confluence pages when the key is getting context into your LLM
All my problems are my own making
Shoulda known it was a misconfigured env
Why can't I just do things right the first time?
Lowkey wanna have a drone drop a meshtastic solar node on the top of my office building
Qwen3 release is massive
Models that you can actually run locally, on par with SOTA models
12 hour eta to download the latest qwen models wtf
When 3 waiting room
Perfect timing for the MoE rig
If you're comfortably cash flow positive is there any reason to take MSTY over MSTR?
MSTY says it's limited upside potential in exchange for cash flow
So if cash flow isn't an issue are you better off in MSTR?
Llama4 here we come - 512gb RAM
Too tired to swap in the 3090 and big psu tonight
Too tired to swap in the 3090 and big psu tonight5 hour drive please shill any cool audiobooks
Gemini 2.5 flash feels amazing, sub 1 cent instant api calls
Stop treating robots like second class citizens
End the anthro-centrism
The fact is your website is more useful if I can feed it to my LLM
We already pay for an account, let us use it
Localllama is pretty dead these days