91–100 of 153 posts
> this is great ;-) Next I’d like to see competing models roast each other
We need a comedy benchmark now
> Okay, let's tackle this roast based on the user's data. First, I need to pick the three most unusual bits. The user is using Safari 17.6 on Mac OS 10.15.7. Wait, macOS 10.15 is Catalina, which is pretty old now—like from 2019. That's a bit outdated, so that's unusual. Next, the screen resolution is 1024x1366. That's a taller screen than usual. Most common resolutions are wider, so maybe they're using an older iPad…
I got similar results and it was kind of hokey. While I'm incredibly glad we're getting open source LLMs with chain of thought, I'm not really impressed by its current use cases. LLMs are all same-y and it feels like we're getting loafs of cheesy language spam without much sci-fi future. I expect for this kind of tech to have really novel use cases. For it to sit between me and the internet and remove the ads, nuke t…
My bet is that hosted LLMs will have advertising baked in and also understand you in a much deeper way so as to manipulate you even more effectively.
Earlier quoted context omitted.
That's a false equivalence. There's more nuance than that. Both OpenAI and Google have a genuine private-sector business model, whereas my claim is that DeepSeek is totally a China Gov't entity.
Why is it an interesting distinction whether a product is privately funded vs tax payer funded? To me it sounds like sour grapes.
For me it just returns a wall of text with (part of) its reasoning about composing a short roast, but not the roast itself. I guess this is not intended.
Earlier quoted context omitted.
I got similar results and it was kind of hokey. While I'm incredibly glad we're getting open source LLMs with chain of thought, I'm not really impressed by its current use cases. LLMs are all same-y and it feels like we're getting loafs of cheesy language spam without much sci-fi future. I expect for this kind of tech to have really novel use cases. For it to sit between me and the internet and remove the ads, nuke t…
I've been working on exactly this. Great to know there's some demand for it. It's now my daily driver for web access. It monitors for content I'm interested in (that's how I found your comment), handles all my searches and feeds, can dynamically adapt its interface, and is working on integrations to submit content for me so that I don't have to leave that interface to write these replies. Hoping to release it early M…
Is there a GitHub or Discord or anything where I subscribe to the project?
For me it just returns a wall of text with (part of) its reasoning about composing a short roast, but not the roast itself. I guess this is not intended.
This ended up impacting quite a few users; the response is streamed from a Vercel function where I had not changed the default timeout, so at 10 seconds it was just cutting things off where it was. Some invocations of the DeepSeek API take much longer than others. This should be fixed now with a longer timeout.
Earlier quoted context omitted.
This ended up impacting quite a few users; the response is streamed from a Vercel function where I had not changed the default timeout, so at 10 seconds it was just cutting things off where it was. Some invocations of the DeepSeek API take much longer than others. This should be fixed now with a longer timeout.
API keeps going through periods of 1-2 minutes of everything timing out... then works well for ~5-10 minutes. Might be better to host this model yourself for serious use right now.