⚡️ nedzen.eth retweeted
Claude is better than you think when it comes to designing real objects. Spent most of my time drilling hoes on the wall than designing/making this thing.
The way I work nowdays, is to a large extend just explaining the AI what I want to do and when I go to Figma is just to make stuff like this. The actual design moved from Figma into a Skills + design system.
2
1
39
What I mean by that, in a real case scenario where I change a button's color the tool, knowing the context of the change would either edit primary color value, or add a new token or ask how to what extend I want to propagate the change.
1
6
I ditched all the (monthly) saas subscriptions, only kept tuta, opencode and x. I think it would be amazing if more saas would switch to a ussage based pricing, they'd keep more customers.
3
45
Keeping a clear mind to act from a place of clarity takes some effort and stepping back from the stream. It is time to meditate
36
⚡️ nedzen.eth retweeted
You say a few magic words to an AI and a physical thing appears at your door in a few hours. 3d printing? CAD? DFM? Who cares? I just need the thing.
You think you know how to prompt but, then you read this article from @felixrieseberg and your whole paradigm shifts 🪄
felixrieseberg.com/how-to-be…
39
btw, kimi-k3.1-pro is available on opencode go with a generous amount of usage. I'm using it in omp, gave it a high altitude task (considering outcome allowing freedom on the how) and it immediately spun 4 agents with deepskeek flash, completed the plan then delegated the execution.
39
卧槽,这踏马太毒了
DeepMind 掌门人、诺奖得主 Demis Hassabis 刚刚把延续了上百年的现代学校教育底裤彻底扒光了:
全球几十万名数学老师每天关在办公室里,为同一门中学数学课重复手搓着一模一样的 PPT,这简直是人类智力资源最荒诞的工业级空转。
很多人至今还把名校名师的课堂讲授当成稀缺资源,拼了命去抢学区房和重点班,
但 Hassabis 直接用最冷酷的第一性原理砸碎了这个幻觉,
标准化的知识灌输从来不配占用昂贵的线下课堂,死记硬背和刷题完全应该被踢出教室,交给能按每个孩子毫秒级反应速度实时调频的私人 AI,加上全世界最顶尖的那几个大师录好底层大课,就足以把全球基础教育的讲授环节一次性彻底平替。
过去两百年里,学校本质上是一座把千差万别的碳基大脑强行塞进同一条流水线、按同一个节拍器拧螺丝的工业时代兵营,
老师沦为了人肉复读机,学生被困在枯燥的题海里比拼谁更像一台劣质硬盘,
而一旦 AI 和顶级大师把标准化的知识传递彻底接管,线下教室唯一剩下的合法存在价值,就只剩下了机器永远无法通过权重计算出来的五个维度:
人际博弈、复杂项目实战、创业野心、创造力,以及最致命的判断力。
这里有一个极度冷酷的反直觉真相:
当个性化 AI 把考高分的边际成本砸穿地板,按标准答案培养出来的做题家,在出校门的那一刻就已经直接沦为了算力废料。
未来的阶层分化,从来不再取决于谁的脑子里背下了更多现成的公式,
而是取决于当 AI 把一万种完美方案同时拍在桌上时,谁拥有拍板做决定的判断力,谁能攒起一个活人团队去把事情干成,
所以别再逼着孩子在死记硬背的泥潭里跟硅基芯片比拼记忆容量了,当流水线式的教书匠和做题家同时迎来历史性失业,谁能先把教室变成训练人性直觉与创业野心的实战沙盘,谁才能在算力时代拿到唯一不被平替的入场券
⚡️ nedzen.eth retweeted
We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 🪄
To enter:
🟢 Share this post
🟢 Comment #RTXON
Terms and Conditions: nvidia.com/en-us/geforce/con…
on a mac mini m4 pro 64 gb ornith 1.5 122B 4-bit works at 10tok /s. quite slow.
I tested flint-int4 too and it actually performed better at 80tok/s
1
1
85
Build a single-file Python 3.11 CLI at /tmp/l4c/ledger.py — no third-party deps. Do not read or copy any other directory.
Requirements:
1. Double-entry ledger stored in ledger.json with atomic writes (temp file + os.replace), safe against two concurrent processes appending at once.
2. Commands: open <account>, post <debit> <credit> <amount_cents> <memo>, balance [--json], history <account> [--limit N], audit.
3. Invariants: amounts are integer cents, never floats; every post must keep sum(debits) == sum(credits) per currency; audit re-derives balances from the journal and reports any mismatch; invalid input exits non-zero with a one-line error and must not corrupt the file.
4. python3 ledger.py selftest runs an in-process test suite covering: concurrent writes (spawn 2 subprocesses), unicode memos, negative/zero amounts, malformed JSON recovery, and the audit invariant. It prints PASS/FAIL per case.
Keep reasoning under 40 lines, then write the complete file in one pass and run selftest. Paste the real output.
43










