prettyblocks

18 ms·

121.
by prettyblocks 2y ago
Same here. I just built a pc with a 3090 for local llm and stable diffusion and have zero regrets.
122.
by prettyblocks 2y ago
I really liked it. I found it to be quite difficult, but somehow managed to solve the problem in 4 tries. At first I didn't quite realize that you can put opposing pieces on the board so I think I wasted one or two turns. In general, l
123.
by prettyblocks 2y ago
The distilled versions I've run through Ollama are absolutely censored and don't even populate the <think></think> section for some of those questions.
124.
by prettyblocks 2y ago
Cool, thanks for the reply. Looking forward to following along with your progress.
125.
by prettyblocks 2y ago
Hey Simon - In your experience, what's the best "small" model for function/tool calling? Of the ones I've tested they seem to return the function call even when it's not needed, which requires all kinds of meta
126.
by prettyblocks 2y ago
This is my intuition as well. DeepSeek and the proliferation of local LLMs have showed us just how much can be done with relatively light hardware. It finally has me wanting to build a rig for local use when before I felt like the technolog
127.
by prettyblocks 2y ago
Have you played with moondream? Pretty cool small vision model that did a good job with bounding boxes when I palyed with it.
128.
by prettyblocks 2y ago
Excellent share - nice to see people doing cool things with the tech while not taking themselves too seriously.
129.
Ask HN: Is anyone doing anything cool with tiny language models?
684 points by prettyblocks 2y ago|346 comments
130.
by prettyblocks 2y ago
It could get really interesting if they allow webhooks and structured output
131.
by prettyblocks 2y ago
I remember when we were in a recession post 9/11 prices were dropping left and right, but right now prices continue to increase and people seem to be paying them. Can we really say that's a notable decline in economic activity?
132.
by prettyblocks 2y ago
I think it's more about the challenge/curiosity of creating the implementation. I'm not sure these AI driven text-RPGs will find long term audiences.
133.
by prettyblocks 2y ago
Most of them hijack search results and do cookie stuffing.
134.
by prettyblocks 2y ago
Not to mention cell phones everywhere.
135.
by prettyblocks 2y ago
I have a similar approach, but the mess can be contained by asking for optimizations and refactors very frequently and only asking for very granular features.
136.
by prettyblocks 2y ago
I prefer vscode+copilot. It's much cheaper and has all the functionality I want. There's access to 3.5 sonnet, and it can edit/create up to 10 files at a time.
137.
by prettyblocks 2y ago
Post-scarcity world means that a large chunk of society has this kind of freedom, so there wouldn't be the same kind of isolation that goes along with currently being one of the few rich people.
138.
by prettyblocks 2y ago
What kind of AI projects are you working on?
139.
by prettyblocks 2y ago
Lots of doomers on here, yet somehow I keep meeting highly technical people still working on interesting things that are in their 40s, 50s, 60s, and older. I think HN probably skews really young and tech tends to fetishize youth, so this is
140.
by prettyblocks 2y ago
> I have a laptop running Linux with core i9 (32threads) CPU, 4090 GPU (16GB VRAM) and 96 GB of RAM. Is there somewhere I can find a computer like this pre-built?