A world model inside a world model 🌏 https://t.co/XQCFzy2pjc
— hardmaru (@hardmaru) 7 août 2025
A world model inside a world model
By
–
A world model inside a world model 🌏 https://t.co/XQCFzy2pjc
— hardmaru (@hardmaru) 7 août 2025
A world model inside a world model
By
–
strange in the social media era to feel that everyone is working on NeurIPS reviews & rebuttals rn but no one is talking about it publicly. i contributed to three rebuttals. one of the three was extremely frustrating. who else is working on rebuttals? how's it going for you?

By
–
Sam is hinting that GPT-5 is a massive world-changing update. What are your predictions? My predictions for GPT-5: Auto-routing between models – 5o (fast multimodal update to 4o) for most tasks, o4/o4-pro for complex reasoning. The system picks for you. OpenAI Coding
By
–
The Synthetic Data Trap: How Your AI Strategy Is Training You To Be Average
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @lexfridman @sama @kaifulee @ID_AA_Carmack @karpathy @2morrowknight @ylecun http://
ow.ly/xhQc30sOj7m
By
–
For training experiments I try to avoid sweeps of all kinds. Instead I aim to build a strong intuition by running targeted experiments one at a time.
By
–
Oh I just use python/fastai. This can help too: https://
fastgpu.fast.ai
By
–
Do you think the synthetic data is relevant here? What I mean is that some effective compute has been embedded into the data to generate these high quality tokens, so you don't need as much compute to get value out of them

By
–
Some quite exciting improvements in genome assembly accuracy from @GoogleResearch and @ucsc Genomics Institute!
By
–
And the whole of http://
course.fast.ai is notebooks.