ac29

4 ms·

1.
by ac29 4d ago
gpt-oss-120b only has 5B active parameters, so its not surprising Qwen3.8 27B outperforms it (Qwen3.8 is also ~13 months newer, which is forever in LLMs)
2.
by ac29 8d ago
I did that with Codex+Sol alone, and made something even more complex in less time with Codex+Astra over the weekend. Its maybe more hands on than you want it to be, but you mostly just need to supervise the work. But yes, the CAD outputs g
3.
by ac29 9d ago
US prices dont include taxes, EU prices do
4.
by ac29 11d ago
GPT-6 is multimodal, LLMs alone have no vision capability
5.
by ac29 16d ago
the methyl groups make it organic chemistry (and dangerous), elemental mercury is fairly safe
6.
by ac29 16d ago
> AMD Strix Halo is about to release 192GB system which is a big upgrade to their current 128GB ai pc Big upgrade to memory capacity but memory speed is only going up by a few percent, so its still going to be slow with more than a few B
7.
by ac29 17d ago
Micropython runs on tons of stuff and while it isnt capital-P Python, its close enough for doing a wide variety of embedded work without learning a new language/libraries/etc.
8.
by ac29 22d ago
Because process affects performance per watt more than architecture so its a useful data point. Even knowing nothing about performance, knowing that it is N6 means it probably isnt particularly power efficient (as compared to leading edge x
9.
by ac29 23d ago
> FCC allows suppliers to self-declare conformance with EMI/EMC rules Yes, though they can also ask for your lab results or to inspect your testing facility and hit you with massive fines if it turns out you falsified your complianc
10.
by ac29 24d ago
Same channel width and power restrictions. The US' 915MHz band is much much better than the EU 869MHz band
11.
by ac29 24d ago
Not sure I trust a benchmark where Haiku gets a nearly perfect score and Fable is tied for last place
12.
by ac29 26d ago
I have Kagi and ChatGPT subs and would agree ChatGPT is the better offering. But you can use Luna with Kagi Assistant on the cheap plan which works pretty well too
13.
by ac29 27d ago
There haven't been any free resets in the past week, there were 4 in the first half of the month
14.
by ac29 29d ago
Cost per task: Model Score Cost / Task Output Tokens / Task ------------------------------------------------------------------------- Muse Spark 1.2 (xhigh) 56.8 $0.40
15.
by ac29 1mo ago
> Using Google products in general is an effing nightmare as soon as you have to give them money Spending money via Google Pay on Android is extremely easy, Google does know how to accept customer's money (in the consumer space)
16.
by ac29 1mo ago
Last time I used the AI studio free tier it was limited to one or two requests, effectively useless. I think people are complaining its too hard to set up a paid API key (no reason you should make paying customers spend more than a few clic
17.
by ac29 1mo ago
Muse Spark 1.2 benchmarks just shy of Opus/Sol and is significantly less expensive than Sonnet (which admittedly is overpriced). Haven't personally used it though
18.
by ac29 1mo ago
China has as least much engineering talent as the US, the claims that Chinese models must just be distilled US models feels like xenophobia. There are more plausible explanations for why the models are similar - all the labs are buying the
19.
by ac29 1mo ago
I was under the impression labs released post-training checkpoints fairly often? So Model N/N+1 might literally have had the exact the same pretraining run and only differ on how much/what kind of postraining they got
20.
by ac29 1mo ago
I haven't had bluetooth issues in years, but I did have an agent reverse engineer a smartphone app that was required for programming some BT headphones. Now I can push my desired runtime settings automatically to the headphones when th
21.
by ac29 1mo ago
Your earlier post is misleading, since "it" sounds like the model, not the installer. (of course if the latter was created by the former, its basically the same complaint, but I suspect what you are talking about was a deliberate
22.
by ac29 1mo ago
Any idea what pricing will be like? The fact that you call out Drone and Defense industries on your page suggests to me this is extremely expensive. I recently priced out getting a PCB done in China and its insanely cheap. My design isnt co
23.
by ac29 1mo ago
One can piss away a lot of money gardening with little to show for it, especially if you buy bagged soil. Buy a raised planter, some soil, some fertilizer, some seedlings, a couple garden tools and you can easily be in for $150+ which is de
24.
by ac29 1mo ago
For reference, IBM was founded in 1911. At the time "computer" was a profession, not a machine, and the majority of US households had no electricity or telephone service
25.
by ac29 1mo ago
Not so much hardware these days, about half of IBM's revenue comes from software (Red Hat related stuff alone probably brings in a ton), and about 20% comes from consulting
26.
by ac29 1mo ago
Doesnt seem suspect to me, training runs have checkpoints and there is no reason you cant release a checkpoint even if you are still training the model
27.
by ac29 1mo ago
> They chose to compare against Open AI’s mid tier model Terra instead of Sol and still lost some benchmark against it. If you scroll very slightly farther there is a benchmark that includes Sol, showing it outperforming Terra (as expect
28.
by ac29 1mo ago
Yeah Anthropic is the only company I am aware of that only allows subscription usage in first party products
29.
by ac29 1mo ago
> Apple has worked very hard to make unified memory a feasible approach, and the benefits of that are pretty clear in apple silicon- that efficiency not only results in power and therefore thermal gains, but also in a significantly faste
30.
by ac29 1mo ago
llama.cpp has used mmap by default for years

More ›