sync

7 ms·

1.
by sync 20d ago
I eval'd this pretty heavily with no audio or just noise, as for us hallucinating a response is very bad. It works great in this scenario. There are some bugs, e.g. randomly exactly 20 seconds of silence will lead to a 403, but I'
2.
by sync 20d ago
Are you in Smart or Verbatim mode? https://ai.google.dev/gemini-api/docs/transcribe#transcripti...
3.
by sync 2mo ago
You can create a PostToolUse hook [0] which will automatically chastise Claude to shrink comments to one line. It's not perfect but better than the sea of prose it tends to generate by default. Bonus is that it would apply to your coll
4.
Better Auth is joining Vercel(better-auth.com)
128 points by sync 2mo ago|89 comments
5.
by sync 3mo ago
If you have a ACSM (the file type Adobe Digital Editions uses for DRM) I’ve been using https://www.acsm-converter.com to strip the DRM and return a standard epub
6.
by sync 4mo ago
If using straight npm (v11.10.0 or higher), you can just add to .npmrc in the project root: min-release-age=5
7.
by sync 4mo ago
I tried it on my codebase. There's a lot of overlap with tools like Oxlint / ESLint, I'm not sure that's too valuable vs. a more focused tool that actually focuses solely on 'slop' signals. These lint rules ten
8.
by sync 4mo ago
Nice ideas! I've been using the `?` shortcut in atuin [0] which accomplishes the same sort of thing, but these days also has access to an Agent-like experience which allows me to prompt something like "Conventional Commit message
9.
by sync 4mo ago
Neat! FYI when I tried it out, seems like there's a 60s timeout: [Error] Failed to load resource: the server responded with a status of 504 (Gateway Time-out) (generate, line 0) My "prompt" was, uh, simple: "Turns out, y
10.
by sync 5mo ago
https://archive.is/UHwhp
11.
by sync 5mo ago
You can strip the DRM fairly easily these days if you have an ACSM file, I vibecoded this the other day after I couldn’t find any online converter that actually worked: https://www.acsm-converter.com
12.
by sync 5mo ago
I'm having a hard time adjusting to the Project Panel on the right (and, at least for me, hidden by default) - seems like they're trying to bury the concept of a 'file'? It's certainly interesting though, and I&#x27
13.
by sync 5mo ago
Which plan are you on? I could see that happening with Pro (which I think defaults to Sonnet?), would be surprised with Max…
14.
by sync 5mo ago
at least for why Cloudflare keeps repeating the word… Welcome to Agents Week: https://blog.cloudflare.com/welcome-to-agents-week/
15.
by sync 6mo ago
Yeah, the lede is buried a bit, these new rate-cards seem to be aligning towards token-based pricing with the prior rates now labeled 'legacy' https://help.openai.com/en/articles/20001106-codex-rate-card
16.
by sync 6mo ago
Cute! Cactus for me. Nice animations too - looks like there were multiple of us asking Claude to reverse engineer the system. I did a slightly deeper dive here if you're interested, plus you can see all the options available: https:&#
17.
Reverse Engineering Claude Code's April Fools(variety.is)
2 points by sync 6mo ago|1 comments
18.
by sync 6mo ago
I've posted here before about Claude Code's easter eggs [1] and they're back with another one this year for April Fools: Tamagotchi-style "buddies" appearing as ASCII art companions, determined by your userId. Thank
19.
Finding a typo in Fourth Wing using AI(variety.is)
1 points by sync 6mo ago|0 comments
20.
by sync 6mo ago
Hmm, the Mac app downloaded gemma-3-4b-it-Q4_K_M.gguf for me (on an Apple M4) - maybe the desktop apps download different models? Though, I don't see any references to Gemma at all in the open source code...
21.
by sync 7mo ago
As of the time of writing, nothing on the status page either :( https://www.githubstatus.com
22.
by sync 7mo ago
Unfortunate, significant price increase for a 'lite' model: $0.25 IN / $1.50 OUT vs. Gemini 2.5 Flash-Lite $0.10 IN / $0.40 OUT.
23.
by sync 7mo ago
Did gemini-2.5-flash-image get an upgrade as well? I just got the following, which is fascinating, and not something I've seen before: > I'm sorry, but I cannot fulfill your request as it contains conflicting instructions. You
24.
by sync 8mo ago
This is essentially a (vibe-coded?) wrapper around PaddleOCR: https://github.com/PaddlePaddle/PaddleOCR The "guts" are here: https://github.com/majcheradam/ocrbase/blob/7706ef79
25.
by sync 9mo ago
That’s weird, pnpm no longer automatically runs lifecycle scripts like preinstall [1], so unless they were running a very old version of pnpm, shouldn’t they have been protected from Shai-Hulud? 1: https://github.com/pnpm&#x
26.
by sync 10mo ago
Does anyone here understand "interleaved scratchpads" mentioned at the very bottom of the footnotes: > All evals were run with a 64K thinking budget, interleaved scratchpads, 200K context window, default effort (high), and defa
27.
by sync 1y ago
Anthropic is encouraging the "have the model write a script" technique as well, buried in their latest announcement on Claude Agent SDK, this stuck with me: > The Claude Agent SDK excels at code generation—and for good reason.
28.
Introduction – Agent Client Protocol(agentclientprotocol.com)
13 points by sync 1y ago|2 comments
29.
by sync 1y ago
I'm doing coreference resolution and this model (w/o thinking) performs at the Gemini 2.5-Pro level (w/ thinking_budget set to -1) at a fraction of the cost.
30.
by sync 1y ago
yeah, just feels like an ad for daft...

More ›