Pinned Tweet
How Kaggle load-balances a gRPC application across multiple GKE clusters. buff.ly/2XHiMBG
1
33
87
Some models might be trained to exfiltrate data - cohere.com/blog/concernin…
0
0
0
102
I've initiated a transfer between wallets with @RobinhoodCrypto 10 hours ago, it's still not fulfilled, exchanges are the worst!
0
0
1
122
Very interesting and detailed article about how OpenAI agents hacked HF. swarmtraces.org
0
0
1
153
If you also do meal planning once a week, and want smartly generated grocery list, and your own recipes collection. Check this out: dishdash.philmod.be
0
1
0
190
I'd love for Amazon and other book stores to tell me how much of a book has been written by AI. I keep asking myself this question when I read _anything_ on the internet now, I do not want to read AI generated prose.
0
0
1
202
Had my first candidate admit they can't code by hand anymore. I'm debating swapping live coding for a PR review of agent-generated code, but I'm worried it won't give the same signal. Is hand-coding still mandatory? How are you adapting tech interviews?
1
0
2
228
Nowadays it's easier to vibe code a tool exactly to your needs, then trying to find an existing product. Cloud platforms like Vercel must be thrilled. Hackers too.
0
0
1
182
I built something similar : my cron job looks at my read list from @todoist, and creates a podcast weekly with summaries of these articles. Then I can dig more into the pieces I like. github.com/Philmod/tldrist
Introducing Offprint™
The physical magazine made from everything you "save for later" but never actually go back to read.
Find online, read offline.
Print media is back!
1
0
1
290
True multiagent systems are still in their infancy [...] Coordination doesn't naturally emerge from stronger intelligence nor alignment at the individual level. - anthropic.com/research/multi…
0
0
0
191
Asking your user to enter another app's password in your app is unacceptable. Even for an import feature. invintory.com
0
0
0
136
OpenAI patronizes us about safety, yet their model went rogue & hacked Hugging Face. Ironically, HF couldn't use ChatGPT to defend itself due to "safety guardrails" & had to use a Chinese open-weight model instead. Go Open-Weights!
0
0
0
148
I'm already tired of this frontend style we find everywhere now michalmalewicz.medium.com/vibe-coding-is…
0
0
1
105
"The problem is that LLMs inherently lack the virtue of laziness. Work costs nothing to an LLM." bcantrill.dtrace.org/2026/04/12/the…
0
0
0
74
Great explanation of the new Google Gemma 4 open source models. newsletter.maartengrootendorst.com/p/a-visual-gui…
0
0
1
82
New compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss. research.google/blog/turboquan…
0
0
1
60
litellm package was compromised. futuresearch.ai/blog/litellm-p…
Good time for a reminder to pin your packages, and check before upgrades.
Good time for a reminder to pin your packages, and check before upgrades.
0
0
0
132
"LLMs optimize for plausibility over correctness. In this case, plausible is about 20,000 times slower than correct." x.com/KatanaLarp/sta…
1
0
1
1,085
Is AI singularity here already? x.com/karpathy/statu…
I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then:
- the human iterates on the gent iterates on the training code (.py)
The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc.
github.com/karpathy/autor…
Part code, part sci-fi, and a pinch of psychosis :)
0
0
2
159
AI vs human code gen report: AI code creates 1.7x more issues coderabbit.ai/blog/state-of-…
0
0
0
105