srcreigh

6 ms·

1.
by srcreigh 4d ago
LLMs can help people understand mathematical principles too. The idea that when LLMs produce solutions, people won’t try to understand them and won’t learn from it, is obviously not true. Terry Tao himself spent time digesting and simplifyi
2.
by srcreigh 4d ago
Yeah. This makes sense. Someone who can dedicate a lot of time to scaling a SaaS serving many customers are helped a lot by AI. I believe busy professionals who already have successful businesses won't have the focus to build a one-off
3.
by srcreigh 4d ago
No, this cannot be the actual issue. Most math textbooks have solutions in the back. That didn't wreck peoples ability to learn math, did it? So we are left to speculate about why solutions in appendix are fine, but LLM solving open pr
4.
by srcreigh 4d ago
At least for now, not everybody can make a private SaaS. Doing that requires skills and focus. Even with SOTA models, as of ~today, there are inflection points where the AI would not be able to make the right choice without a knowledgable h
5.
by srcreigh 5d ago
Without pinning providers, you pay let’s say 2x more for agentic coding, since that many input tokens aren’t cached as you bounce between providers.
6.
by srcreigh 8d ago
Math can never be fully solved by a computer, if only for lack of computational resources.
7.
by srcreigh 8d ago
The OCA (Orthodox Church of America) has a hagiography page regarding them. https://www.oca.org/saints/lives/2014/11/19/100292-saints-ba...
8.
by srcreigh 13d ago
How many years until chips like this are available to consumers?
9.
by srcreigh 13d ago
It is 15x more expensive. Openrouter usually charges like 1/4 for cached input. Most of the cost for agentic coding is input tokens, you pay for the whole context at each tool call or message. Output tokens is just a small rate
10.
by srcreigh 13d ago
Great observation. That’s not enough context even for some one shot xhigh requests. When I put Qwen3.8 27B xhigh towards adding scope proxying to the Guice library, it one shotted a great impl using 250k context before stopping. Part of the
11.
by srcreigh 13d ago
No, I don't own any of this hardware, im just summarizing what I could find by reading others posted benchmarks
12.
by srcreigh 14d ago
There is comparison actually. I spent all day researching this a few days ago. Memory-wise, the RTX PRO 6000 can barely hold two 1M context Qwen 3.8 27B models at 8 bit quantization at the same time. The 512GB M5 Ultra Mac Studio could hold
13.
by srcreigh 14d ago
Makes you realize how insane the M5 Ultra Mac Studio is. 1.2TB/s bandwidth 512GB memory. Its rated max power draw is just 480W. And it also has amazing M-series CPUs. It costs less than just one of these GPUs which each take 700W to ru
14.
by srcreigh 14d ago
The adults brainwashed us https://www.ikea.com/ca/en/p/barndroem-box-beige-70560615/ https://www.ikea.com/ca/en/p/vallaby-rug-green-10548216/
15.
by srcreigh 14d ago
Open weight models have been getting better/smaller every year. Also, from what I can tell, MLX inference is not as well optimized as CUDA, and the M5 Ultra has additional kinds of AI compute which is unavailable on other M models. Wit
16.
by srcreigh 15d ago
So I ran the math, assuming the agent takes 75 turns per 200k context, with deepseek v4 flash it costs around $2.57 to reach 1M context in 375 turns. Cached input costs scale quadratically with # of agent turns. Considering that I hit the 1
17.
by srcreigh 15d ago
It’s not so clear after 5 years that you’ll come out ahead. You’ll have spent $20k. The apple computer owner will probably be running local models that are better than today’s frontier on the same hardware. Idk where you live, but where I a
18.
by srcreigh 15d ago
They're not cheap at all. I did one xhigh Qwen 3.8 27B agentic coding task last week via OpenRouter and it cost me like $10. 99% of the cost was in input tokens, I only used like 100k ish output tokens. It was a one shot task asking th
19.
by srcreigh 18d ago
Well, yes, but if k is for example Ω(n) then O(n-k) is also O(1).
20.
by srcreigh 20d ago
Do any other aphantasia folks have the experience of like, that one time they saw a very vivid image, but it only ever happened once for a short time? That happened to me a few years ago. It was a very interesting experience. It’s surefire
21.
by srcreigh 20d ago
No. The bitter lesson is about capabilities. GP is talking about efficiency. GP isn’t suggesting that focused narrow model(s) will be more capable than large model, but that many small focused models can have sufficient capability while bei
22.
by srcreigh 24d ago
How should I learn more about how to do it, what to buy, etc ? I haven't found ChatGPT to be a good teacher about this topic, and in particular re glitching, AI will refuse to discuss specifics I have enough basic soldering to get UART
23.
by srcreigh 24d ago
> I haven’t actually been brave enough to write a modified firmware to the thing yet - it’s a pretty expensive monitor - but I’ll get there at some point. Honestly if you don't have working patches, it's really not owned. I wou
24.
by srcreigh 25d ago
Side note, does anybody know of some good Rhombus codebases I can read?
25.
by srcreigh 26d ago
Note M1 and M2 macs don’t have hardware support for nested virt.
26.
by srcreigh 29d ago
Re Forgejo actions, I'm just toying around. Their v15 release (apr 16) really enabled k8s native runners. They added an ephemeral runner API and a bunch of APIs to get jobs. That's what im using to do k8s autoscaling
27.
by srcreigh 1mo ago
Yeah. MicroVMs make hosting runners pretty straightforward. I had codex make a forgejo runner controller which pulls forgejo actions and creates kubernetes jobs for them. It was a bit of a pain to configure firecracker with k3s. It really c
28.
by srcreigh 1mo ago
It’s fascinating that in order to do this, they had to remove 50% of reads and 33% of transactions from the main DB.
29.
by srcreigh 2mo ago
An impressive number of things don’t make sense about this thought experiment. Whatever the genie says wouldn’t have much to do with the life you live afterwards, since your life changes after hearing the info. One persons careers worth of
30.
by srcreigh 2mo ago
Pages (especially threads while logged in) taking seconds to load definitely happened to me pre SQLite migration.

More ›