ac29

4 ms·

31.
by ac29 2mo 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
32.
by ac29 2mo ago
Yeah Anthropic is the only company I am aware of that only allows subscription usage in first party products
33.
by ac29 2mo 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
34.
by ac29 2mo ago
llama.cpp has used mmap by default for years
35.
by ac29 2mo ago
The article notes Google Cloud revenue grew 82% YoY
36.
by ac29 2mo ago
I took an embedded C/C++ class online 5-10 years ago and hated it. I dont want to be doing bit shifting math to put some device's registers in the correct state for me to write text to a screen
37.
by ac29 2mo ago
Yeah I did a project with ESP32 and micropython recently and getting the proof of concept up was very very quick. The agent I was using also had no problem writing drivers for peripherals that didnt already have micropython drivers (I seem
38.
by ac29 2mo ago
> I’d really rather we didn’t kill animals tbh, but when it comes to growing our food we have little choice right now Several hundred million Indians would disagree
39.
by ac29 2mo ago
Its just a single benchmark, but Luna 5.6 xhigh scores within the margin of error the same as Opus 4.8 max on DeepSWE for 8x cheaper. Luna max is quite a bit higher than Opus and still 4x cheaper
40.
by ac29 2mo ago
OpenCode Go is a great deal but I recently dumped my subscription because I found myself rarely reaching for it over my Anthropic sub (I can get 40 hours of work a week out of the $20 sub and almost never hit weekly limits). Subscribed to O
41.
by ac29 2mo ago
You do understand that the "frontier" people are usually talking about is the cost-intelligence frontier right? By definition there is no model that is both cheaper and as intelligent or better than another on the frontier.
42.
by ac29 2mo ago
> You will see people claim Claude uses 2x to 4x the tokens of GPT. My guess is something to this effect was in the prompt and the LLM made a point of correcting it
43.
by ac29 3mo ago
Try an 8 bit quant of Qwen 35B, but temper your expectations. Those Qwen 3.6 models are impressive for the size, but you need an order of magnitude more parameters to actually be useful for more than trivial work in my opinion.
44.
by ac29 3mo ago
Phased array radars are export controlled in the US. It doesnt mean its illegal to build or own, but it might be illegal to sell in some cases
45.
by ac29 3mo ago
llama.cpp supports a wide variety of 4-bit and smaller quants and mmap's models by default, so you dont need to be able to hold the weights in memory (the OS will handle bringing them in from storage as needed) Its cool to see this imp
46.
by ac29 3mo ago
> So... maybe we can still use third party harnesses with Claude Code subscriptions... for now? The way I read this is: yes, if the third party harness uses Anthropic's Agent SDK. Most of them do not, AFAIK, and are still against
47.
by ac29 3mo ago
I have a SwitchBot Meter Pro CO2 that is just temp/humidity/C02. Works well and was inexpensive ($50 or so).
48.
by ac29 3mo ago
I've got a NDIR CO2 sensor about 6 weeks into a supposed 1 year battery life, still reporting 100% battery. Power use is very low since it only fires up the CO2 sensor every 30 mins.
49.
by ac29 3mo ago
Giving users ability to use apt with root privileges is pretty much game over security wise. Full root is a malicious package away
50.
by ac29 3mo ago
> Anthropic just provides a subscription - which Enterprise usually doesn't want you to use because everything you're submitting through that will be trained on / becomes part of their model. My Pro account very clearly ha
51.
by ac29 3mo ago
The inference providers are running batch sizes much larger than 10
52.
by ac29 3mo ago
I'm certainly willing to believe the M1 was the most energy efficient, but your data center didnt have anything faster than a laptop?
53.
by ac29 3mo ago
I did my first ESP32 project recently and was amazed you can get a system that starts up Micropython, then a Wifi AP, DNS, and Web Server in a second or two total and uses less than 512kB RAM. And thats with a high level programming languag
54.
by ac29 3mo ago
> IBM no longer owns any fabs Per IBM: "IBM Research at Albany [...] includes more than 100,000 square feet of semiconductor fabrication space" I guess that is technically a R&D fab not a production one, but they definitely
55.
by ac29 3mo ago
Most real financial transactions need to start and end in fiat though, and that's where the costs are
56.
by ac29 3mo ago
A change in regulations is not a meritless argument, it's useful information
57.
by ac29 3mo ago
You could subscribe to Anthropic/OpenAI for the rest of your life for the cost it would take to host GLM5.2 locally - you need 1.5TB of VRAM just for the weights
58.
by ac29 3mo ago
> I'd rather use Sonnet than Qwen I get this, though the pace of Chinese releases is relentless. Qwen3.7 Plus/Max (closed variants) feel notably better than Qwen3.6, and Minimax M3 is a big jump from 2.7 in capability as well.
59.
by ac29 3mo ago
If you use uv, python apps use a shared cache which helps a lot.
60.
by ac29 4mo ago
All current Qwen 3.7 models are closed though they have said more releases are coming

More ›