Juan
@KeyboardSloth
Followers
201
Following
2K
Media
160
Statuses
2K
⚙️ Swift Configuration 1.0 is here! 🎉 No more juggling environment variables, config files, and command-line flags separately. One unified, type-safe API to read configuration from any source, and swap providers without refactoring your code. More here... https://t.co/6O1K1AxTv4
3
57
382
Today we started rolling out SimGym — a system that creates “digital customers” that behave like real ones. They browse your site, complete tasks, and reveal optimization opportunities. You can even run A/B tests with *zero* live traffic! Spent a year developing it.
121
166
3K
The consumer GPU fine-tuning ecosystem is about to explode. The efficiency wall just got lower. 3x training speedups mean we're moving from "who can afford training" to "who can iterate fastest." This matters for the LoRA-to-full-training transition. If you can train 4B
You can now train LLMs 3× faster with no accuracy loss, via our new RoPE and MLP kernels. Our Triton kernels plus smart auto packing delivers ~3× faster training & 30% less VRAM vs optimized FA3 setups. Train Qwen3-4B 3x faster on just 3.9GB VRAM. Blog: https://t.co/j3HpsmzrCw
12
45
464
A $0.15 amino acid just outperformed prescription sleep meds. 3g before bed = better cognition & alertness than 8hrs normal sleep. Double-blind trial. No side effects. Here's what 99% of doctors don't tell you:
68
335
2K
WOW! @AnthropicAI released interviews with 1,250 professionals about how they use AI for work. You can find it on @huggingface as an open dataset!
We’re launching Anthropic Interviewer, a new tool to help us understand people’s perspectives on AI. It’s now available at https://t.co/W8P36sPQBy for a week-long pilot.
15
89
994
This Opus 4.5 system card has detailed information about this crazy model https://t.co/Sy8gFLNfsY
12
52
722
Ilya and I are predicting the same future. @ilyasut predicts that as AI becomes viscerally powerful, humans will change in unprecedented ways. I’ve been arguing the same. We are not open to radical change yet because we cannot feel the pressure of AI. Once we do, the vectors for
262
353
4K
running 24x claude code opus's in parallel and it works flawlessly using github as the coordination layer for code reviews, CI checks, and planning
213
69
2K
Database System development will make you realise that 1 millisecond is actually a really really really long time ~300 tx/ms TigerBeetle vs 1 tx/ms PostgreSQL https://t.co/m4WbFfEPzO
44
259
4K
Pro tip for Claude Code: Add this line to your system prompt: "Your context window will be automatically compacted as it approaches its limit. Never stop tasks early due to token budget concerns. Always complete tasks fully, even if the end of your budget is approaching."
platform.claude.com
Claude API Documentation
80
322
4K
You can generate SSH keys on the secure enclave of your Mac, and use that to connect to your servers. Since the OS can’t read any data on the secure enclave, it’s much harder for the keys to get stolen. When you need to use the key, the system will perform biometric
23
68
919
I built a tool to instantly start up a @Cloudflare tunnel to a localhost port using a custom domain (in a zone that you own) It's now open source.
2
0
0
We optimized Claude Code's system prompt - just its prompt - and achieved +10% boost on SWE Bench. No architecture changes, tool improvements, or fine-tuning. Last time we optimized Cline, seeing 15% boosts on SWE Bench, bringing GPT-4.1's accuracy up to Sonnet 4-5, just through
43
109
2K
Just published an article on @huggingface's blog about AnyLanguageModel that goes into more detail about the motivation behind the library and what we're building toward: https://t.co/ctRmsaT8b6 🆕 Be sure to check out our new chat-ui-swift example project (linked at the end)
huggingface.co
3
33
141
NASA engineers have developed a method to extinguish fires using low-frequency sound waves. [🎞️ sambentley] https://t.co/UViSFsPRK1
291
3K
15K