deivid

5 ms·

1.
by deivid 9d ago
Nice! I also got tired of the Withings app (which now pushes AI subscriptions, wtf) and "made" an app for the smartwatches (at least, it works with my ScanWatch 2) The app is at https://github.com/DavidVentura&#x2F
2.
by deivid 17d ago
Kokoro can be hacked a bit (see script https://github.com/DavidVentura/translator-rs/blob/master/sc... ) which makes it about 3x faster. My phone went from slightly slower than realtime to very comfortab
3.
by deivid 23d ago
PaddlePaddle (v6) is fantastic and fast
4.
by deivid 2mo ago
Love the idea, but I can't handle the big screens on these smart watches. To me it feels like having a fisher price toy on my wrist. Hopefully a more open alternative to Withings will emerge
5.
by deivid 2mo ago
Depending on what data you want, you can try Withings watches. I'm currently reversing the protocol and it mostly works.
6.
by deivid 2mo ago
For me, sloppy content is worse than no content. I'd rather have 3 sentences written by a human than a wall of Claudespeak. Do you have an RSS or a way to subscribe to the TTS work you are doing? Your newsletter button is just a mailto
7.
by deivid 2mo ago
It's supposed to handle acceleration though, if you bring your phone to your face quickly dots get larger. I've done some in-car testing, but things get squeamish quickly so it's not the easiest to validate. Changing the colo
8.
by deivid 2mo ago
Interesting project, but the slop readme made me quit reading halfway
9.
by deivid 2mo ago
send me an email
10.
by deivid 2mo ago
send me an email
11.
by deivid 2mo ago
I spent a day fiddling with AI and dropping the expensive layers in kokoro, on phones, on CPU, on MNN, it runs 3x faster. Quality is very close. Will vary in your setup, but here is my script: https://github.com/DavidVentura
12.
by deivid 3mo ago
I am making (open) finetunes for malayalam and kannada (and bengali, gujarati, hebrew), and need someone to transcribe a few images for me. Could you contact me if you are interested in helping?
13.
by deivid 4mo ago
Amazing! I'd recommend using PaddlePaddle OCR models if you want to make your handwriting conversion local, it's definitely fast enough. Probably faster than the Sonnet call. I tried the model on your handwriting and it worked gre
14.
by deivid 4mo ago
Commercial or FOSS? I've been researching the mobile side and it's very exciting!
15.
by deivid 4mo ago
Sounds great, but if you din't cave to apple/google (eg: graphene, lineage), models are not built-in. Every app needs to ship their own models, and they are not tiny. Is there a solution for this? I'm currently just making us
16.
by deivid 4mo ago
There's definitely a much smaller, cleaner kubernetes hiding inside the current monstrosity. LB, Services, pods, DNS gets you like 99% of the way there
17.
by deivid 5mo ago
How do you manage to finish things? After a while, for me, the novelty high wears out, and instead it becomes a wall. Some project require something silly, like sending 1 email to be completed, and it becomes an impossible task
18.
by deivid 5mo ago
This captures a lot of how I've been feeling lately. Thanks for sharing
19.
by deivid 5mo ago
With this approach I managed to get to sub-10ms start (to pid1), if you can accept a few constraints there's plenty of room! Though my version was only tested on Linux hosts
20.
by deivid 5mo ago
Have played it a few times, it's really good
21.
by deivid 5mo ago
Working (again) on an offline translator for Android: https://github.com/davidventura/offline-translator This week I added TTS support, which needed multiple inference pipelines, it was not easy to find models for 50 l
22.
by deivid 6mo ago
Niiiiiice, I've been working on something like this, but reducing linux boot time instead of snapshot restore time; obviously my solution doesn't work for heavy runtimes
23.
by deivid 6mo ago
Says fast, but how fast? Didn't really see any measurements
24.
by deivid 7mo ago
TCC is fantastic! Very hackable, easy to compile to WASM for some interesting in-browser compilation
25.
by deivid 8mo ago
That makes sense, I guess this was not a problem for the times I needed nolibc. I do agree that trying to extract data/logic from linux is a pain -- I've tried a few times to extract some of the eBPF verifier handling, but end up
26.
by deivid 8mo ago
Unikraft is cool, I still have it in my 'todo' list to play around with sometime. Linking the app with the 'kernel' seems pretty nice, would be cool to see what that looks like for a virtio-only environment. Just wanted
27.
by deivid 8mo ago
Would it be cheating to use the kernel's nolibc?
28.
by deivid 8mo ago
You can do <10ms. I was working to see if I could get it under 1ms, but my best was 3.5ms
29.
by deivid 8mo ago
You can boot a vm without snapshots in < 10ms, just need a minimal kernel.
30.
by deivid 8mo ago
This is really well written, thanks for sharing. I didn't understand the point of using Unikraft though, if you can boot linux in much less than 150ms, with a far less exotic environment

More ›