Jacob

@Tu7uruu How many hours are needed to fine tune new language?

Just added in smol-audio: a reproducible NeuTTS Nano recipe for new languages. Want a TTS model in your own language? - NeuCodec dataset encoding - fine-tuning CLI + notebook - Inference script Change the phonemizer + dataset and train your own TTS model. Italian example includes a 300k samples/around 1000 hours of audio YODAS/Granary config.

@saararbel1 הדבר האחרון שמעניין אותנו זה לקום באמצע הלילה לדבג תקלה, מה שמעניין זה להתקדם כמה שיותר מהר. יש לנו מפתחים חזקים מאוד אז לא משנה איזה קוד תתן להם יחד עם קלוד יסתדרו (גם אני)

@eaglescode מעניין, איך אתם שומרים על mental model אחיד של הארכיטקטורה? כמה מהמפתחים באמת מסוגלים לדבאג תקלה באמצע הלילה בלי לבקש מהמודל להסביר להם את הקוד של עצמם ולהיכנס לפאניקה?

אנחנו בדרך למשבר tech debt מטורף וכל הקרסרים והקלודים רק מאיצים אותו אל הקיר

@saararbel1 עם צוות של 10 מפתחים בערך כן, ומערכת מורכבת יש הרבה משתמשים

@eaglescode עובד לך במערכות מורכבות עם צוות מפתחים גדול?

@OpenWeb360 @DanielSmidstrup LOL that is accurate

@DanielSmidstrup USA builds China copies fast Europe regulates

USA has ChatGPT USA has Grok USA has Claude USA has Gemini China has DeepSeek China has Qwen China has Kimi China has MiniMax Europe has?

@DanielSmidstrup Europe has regulations!

@thsottiaux Broo I wanted to take some time to rest 😉

Codex usage limits have now been reset across all paid plans. Enjoy the weekend!

@0xglitchbyte Move fast and break things, the time is short and the work is endless

When are we going to stop pretending oneshotting bun’s rewrite was a good idea?

@eaglebuildz @dreamsofcode_io Yeah, just vibe-code unsafe Rust and expect segfaults

@dreamsofcode_io hot take: code doesn't need to be readable and clean anymore with agentic coding. new paradigm: - just vibe-shit the code - have enough tests written by ai - have enough security checks done by ai - make manual tests nobody needs to read the code anymore.

The code is so bad though. Like the worst Rust code I’ve ever seen. Honestly it’s so bad I’m thinking of fully hand writing code again.

@k_flowstate Java

Name that programming language you are not even bothered to learn in your entire lifetime

/goal research for 24 hours on new methods and architectures for my model. Convert papers from various fields into text, combine ideas across domains, validate hypotheses with uv (without full training) and save everything into experiments.md and summary.md

@DanielSmidstrup Exactly the oppsite

Claude Opus 4.7 is best for coding GPT 5.5 for design front-end Agree?

@mitchellh The time savings are just worth it.

I strongly believe there are entire companies right now under heavy AI psychosis and its impossible to have rational conversations about it with them. I can't name any specific people because they include personal friends I deeply respect, but I worry about how this plays out. I lived through the great MTBF vs MTTR (mean-time-between-failure vs. mean-time-to-recovery) reckoning of infrastructure during the transition to cloud and cloud automation. All those arguments are rearing their ugly heads again but now its... the whole software development industry (maybe the whole world, really). It's frightening, because the psychosis folks operate under an almost absolute "MTTR is all you need" mentality: "its fine to ship bugs because the agents will fix them so quickly and at a scale humans can't do!" We learned in infrastructure that MTTR is great but you can't yeet resilient systems entirely. The main issue is I don't even know how to bring this up to people I know personally, because bringing this topic up leads to immediately dismissals like "no no, it has full test coverage" or "bug reports are going down" or something, which just don't paint the whole picture. We already learned this lesson once in infrastructure: you can automate yourself into a very resilient catastrophe machine. Systems can appear healthy by local metrics while globally becoming incomprehensible. Bug reports can go down while latent risk explodes. Test coverage can rise while semantic understanding falls. Changes happens so fast that nobody notices the underlying architecture decaying. I worry.

@thsottiaux The remote computer use works like magic keep it up 🚢

Codex everywhere

@rezoundous Because they were too late.

50% increase in Claude Code weekly limit doesn't excite me like it used to.

@rezoundous I use low thinking with fast mode on always, otherwise it's too slow (almost unustable IMO)

Am I the only one using GPT-5.5 on xhigh all the time?

@willdepue A great way to break your usb-c port

Tired of holding your laptop half open to keep your agents running? Introducing AgentPlug: A USB-C dummy plug that keeps your Mac in clamshell mode by pretending to be an external display! No commands, no security worries (just pull it out to stop!), no hassle.

@charliermarsh Running an experiment where I do NOT restructure the project with uv (it already uses uv instead of requirements.txt) 😉

Running an experiment where I do NOT rewrite uv in Rust (it is already written in Rust)

@themishra4402 Yes, we don't need much smarter agent, we need more affordable agents, they already gets the job done

After Claude’s next update… will GPT-5.5 still live up to the hype?

@bridgemindai Yea for frontend design it's really good

is anyone seriously using this garbage?