The hardest part of owning the iPhone Duo so far has been rewiring my brain based around this more versatile device.
It's so engrained in my mind to pull out my laptop for "large screen" tasks, to only use a single screen and just deal with it for awkward tasks (like multitasking), and to just accept that typing on a phone isn't that great.
Now with the Duo I can open it up for "large screen" tasks, multitasking, or to get a much bigger and easier-to-type-on keyboard.
I'm sure these are all things most foldable users have learned over the past seven years, but they're new to me as a first-time foldable user.
It's been exhausting, but I'm enjoying it so far.
I have a strong belief that in 10-20 years society will frown upon energy drinks, and really any high caffeine drink, the same way we frown upon smoking today. It will be interesting to see if a company built on the foundation of caffeine like Starbucks will be able to pivot.
Either @AnthropicAI was lying about Fable 5.1 being cheaper and more efficient, or there's a serious bug causing excess token usage. This is less than 24 hours of Fable 5.1 (low) use. Last week I used a mix of about 20% Fable 5 (high) and 80% Opus 5 (high) which lasted the week.
This is on Claude Max 20x by the way. Would love to hear from @claudeai or @ClaudeDevs instead of all this radio silence.
Building an app with AI is addictive as hell. Just one more prompt before bed... just one more... just one more...
Haas livery designers are absolutely unmatched.
Unveiling the King of Monsters 🐉 Our official livery for the Japanese GP takes the iconic @Godzilla_Toho to Suzuka 🇯🇵
#HaasF1 #F1
What is with people letting FSD perform illegal actions and then blaming the system? If it's about to do something illegal you:
1) Disengage the system.
2) Proceed with completing a legal action.
3) Report the problem with a voice note.
You made the wrong choice, unfortunately. grok.com/share/bGVnYWN5LWNvc…
When we turn the clocks ahead on March 8, it will be the last time change ever for BC. We're changing to a permanent daylight saving time, simply called Pacific Time (PT).
Learn more: News.gov.bc.ca/33415
ALT New in BC Permanent Daylight Saving Time Starts Sunday, March 8
Overfitting. He was likely overfitting the model for one specific benchmark.
PewDiePie didn’t “train his own LLM.” He fine-tuned an existing open-source model on coding benchmarks. His model started at 8%, crawled to 16% after format fixes, and one run hit 19.6% that briefly passed GPT-4o on a single benchmark before he couldn’t consistently reproduce it.
The tweet makes it sound like a YouTuber casually built a frontier lab in his bedroom. What actually happened is more interesting: a guy with a $41,000 home rig of 10 GPUs and 424GB of VRAM spent months failing, retraining, and iterating on dataset quality until he squeezed marginal gains out of a fine-tune.
This is the part worth paying attention to. The entire arc from October 2025 to now tells you where AI tooling has actually landed. PewDiePie went from building his first PC to running Qwen 235B locally, vibe-coding a custom chat UI, orchestrating multi-agent voting systems, and now fine-tuning models on custom datasets. He did most of this through AI-assisted coding itself.
The video is literally called “I wish I never did this project.” He’s documenting how painful and tedious the process was. That honesty is the signal. The hype accounts strip that away and replace it with “what the f*ck, YouTuber beats DeepSeek.”
The real takeaway: fine-tuning on specific benchmarks with curated data can let anyone temporarily spike a score past models that cost hundreds of millions to train. That tells you everything about how narrow benchmark gaming has become, and nothing about general capability. PewDiePie knows this. The people quote-tweeting him with shock emojis do not.