tgtweak

15 ms·

211.
by tgtweak 1y ago
I think it's only pulling the older model - I see it's using the liteRT models from May.
212.
by tgtweak 1y ago
Ah, I already had edge installed and it had gemma 3n-e4b downloaded... is this the same model that was previously released?
213.
by tgtweak 1y ago
Any readily-available APKs for testing this on Android?
214.
by tgtweak 1y ago
I keep seeing these "grid synchronizing" inverters that don't require transfer switches and can generate to offset the energy pulled from the grid - in the event of a full grid outage, you can always manually hit the disconne
215.
by tgtweak 1y ago
My experience with heavily quantized models is they do better than a similar sized unquantized model but don't really perform anywhere near the pre-quantized model.
216.
by tgtweak 1y ago
The price seems very high for the components. The top airthings with pm2.5, pm1, CO2, radon, humidity, temperature, pressure and full WiFi + cloud connectivity is $250.
217.
by tgtweak 1y ago
I've been using this for years - amazing while traveling. I wish they'd add some non-bird noises to it as a courtesy since there are often whips and calls that are non birds, like insects or mammals. I think it would add some dep
218.
by tgtweak 1y ago
Took a picture of a bag of chips and it said it was seasoning... I think, for the size, it's OK - but not really there. I'm not sure how they got ELO anywhere near Claude or Gemini... those models are leagues ahead in terms of on
219.
by tgtweak 1y ago
It's a matter of time before we get a limited activation model for mobile - the main constraint is the raw model size, more than the memory usage. A 4B-A1B should be considerably faster on mobile though, for an equivalent size model (
220.
by tgtweak 1y ago
What tooling did you use to make the agents cross-collaborate?
221.
by tgtweak 1y ago
Been using cline and their snapshot/rewind/remove context (even out-of-order) features are really shining especially with larger projects and larger features+changes becoming more commonplace with stronger LLMs. I would recommend
222.
by tgtweak 1y ago
I get the high performance component, but it seems most of their speed is from underlying models supporting a "diff style" edits and that a small percentage of it is from code-level text-editing optimizations and improvements. vsc
223.
by tgtweak 1y ago
You can see it in the cline repo which does prompt based tooling, with Claude and several other models.
224.
by tgtweak 1y ago
The print speeds have also 5x'd with these new high speed printers and it's great for experimentation and prototyping. What used to be a 16-20 hour print on a kit based bed slinger is now 3-4 hours with similar quality.
225.
by tgtweak 1y ago
These are excellent tips. Also helpful is the fact that since Bambu came into the scene, the tooling (slicers) and printers (core-xy and others) are all stepping up their usability game, getting the entire space into a "just print&quo
226.
by tgtweak 1y ago
Yes that is correct, substitute Apple for Samsung or Electron I suppose.
227.
by tgtweak 1y ago
There are two very distinct parts here and the article does a good job of muddying it. Chromium: the base of Chrome, which is opensource - is developed in part by google and many other companies, including Microsoft, Apple, Brave, and dozen
228.
by tgtweak 1y ago
Very nice, should run nicely on a 3090 as well. TY for this. update: wow, it's quite fast - 70-80t/s on LM Studio with a few other applications using GPU.
229.
by tgtweak 1y ago
And vmem use?
230.
by tgtweak 1y ago
Have you tested it on any faster FPGAs? I think Azure has instances with xilinx/AMD accelerators paired. >Standard_NP10s instance, 1x AMD Alveo U250 FPGA (64GB) Would be curious to see how this benchmarks on a faster FGPA since I im
231.
by tgtweak 1y ago
I remember reading that got3.5-turbo instruct was oddly good at chess - would be curious what it outputs as a next two moves here.
232.
by tgtweak 1y ago
It would be great if they published an SDK for canbus communication and interfacing, that would really enable aftermarket and add-ons.
233.
by tgtweak 1y ago
They did have a tendency to shatter in the norther climates though during cold weather. Many of those Saturns in Canada driving around with cracked body panels from stray rocks or errant doors in parking lots.
234.
by tgtweak 1y ago
Did they cook the books on those too? honestly I participated in the IPO then sold when it was up a bit - I had faith in their fundamentals as a business but they did not seem mature as a public company.
235.
by tgtweak 1y ago
I would have put a peltier/TEC below the thermostat to influence it's measured air temp vs resorting to reverse engineering and illegal signal broadcasting.
236.
by tgtweak 2y ago
youtube shorts, same - instagram reels, same - snapchat stories - same. I honestly think the solution has to be external as these platforms cannot govern themselves and are incentivized to keep your eyes on them at every waking moment.
237.
by tgtweak 2y ago
sigh upvote Leave it to k-mart to never let you down.
238.
by tgtweak 2y ago
The frustrating part is they'll give you $1500/day back again if you let them know you're leaving... Very annoying tbh.
239.
by tgtweak 2y ago
Cuda has similar inefficiencies and many use cases can have equal uplifts by going lower level on the code. I think this is what deepseek had done to get their speedups on older hardware. Even way back in the days of GPU crypto mining - cus
240.
by tgtweak 2y ago
The definition of hacking is getting pretty loose. This looks like the sandbox is doing exactly what it's supposed to do and nothing sensitive was exfiltrated...

More ›