inspxtr

maybe port over some of your previous videos to grow content on peertube as well if it’s possible. not sure if there’s any legal issue with this tho.

thanks for the suggestion! will check it out!

gotcha! I’ve just started to use opensnitch (for linux) but I don’t usually inspect the detailed connections that often. thanks for the tip on firefox, I’ll be on the lookout for those.

How do you view these and how would you block them by the way? Via uBlock?

Thanks for the suggestions! I’m actually also looking into llamaindex for more conceptual comparison, though didn’t get to building an app yet.

Any general suggestions for locally hosted LLM with llamaindex by the way? I’m also running into some issues with hallucination. I’m using Ollama with llama2-13b and bge-large-en-v1.5 embedding model.

Anyway, aside from conceptual comparison, I’m also looking for more literal comparison, AFAIK, the choice of embedding model will affect how the similarity will be defined. Most of the current LLM embedding models are usually abstract and the similarity will be conceptual, like “I have 3 large dogs” and “There are three canine that I own” will probably be very similar. Do you know which choice of embedding model I should choose to have it more literal comparison?

That aside, like you indicated, there are some issues. One of it involves length. I hope to find something that can build up to find similar paragraphs iteratively from similar sentences. I can take a stab at coding it up but was just wondering if there are some similar frameworks out there already that I can model after.

yeah agreed with your sentiment. I think it’s good to have an intuition about something, but it’s much better when there’s data to back it up.

Cuz then, they can do the same with others, say Youtube or other streaming services, and start to compare the numbers, like % of ads, what types of ads, how long are the ads relative to content, how many of these ads are political, how many of these ads may be harmful, …

Having these numbers can be quite handy for other researchers and regulators to look into these issues more concretely, rather than just say, “as your brothers and sisters already know, tiktok serves ads”

how bout baserow.io or nocodb cloud? Haven’t used them but I think they’re open source. But they don’t have mobile apps AFAIK for editing.

this is interesting, but it’s not open source yet? Couldn’t find the code. I only saw the author saying that the intent is to be open source.

I think apps like this is really interesting and could really benefit from selfhosting (either/both the LLM or the app deployment), especially due to the potential security/privacy issues, as well as lock-in issues with OpenAI.

while the following is not really my threat model, wouldn’t a person who’s being targeted, say a journalist/activist, have a higher chance of their device being compromised (possibly even physically)? If so, would Session still be a valid option for them?

how so?

feels like someone can build another model to generate texts with these styles.

I don’t get what the obsession with big phones is. Is it that most people really want big phones or that companies can charge more for them?