i like some of Acer’s products, but wtf do they know about global memory constraint
"Supply Shortage Problem Has Disappeared": Acer CEO Says Memory Prices To Decline In 2027 As CXMT Starts Mass Production Of New Chip Platform zerohedge.com/markets/supply…
13
i assume this has been the case for a while
OPENAI HAS LARGELY AUTOMATED TRAINING OF NEW EXPERIMENTAL AI MODELS
OpenAI’s internal AI models can now handle much of the process of building and training experimental models, including writing GPU kernels and optimizing the code used to run them.
Researchers can reportedly give an AI a single example of the optimization they want, then let it work for weeks implementing and testing similar improvements.
OpenAI employees also say internal agents increasingly collaborate with each other to solve problems without involving their human users.
The capability has improved significantly in just the last few months, while OpenAI’s growing access to compute is allowing researchers to test ideas much faster. Employees said some experiments that previously could have taken years can now be carried out in about a week.
Source: The Information
28
couldn’t have said it better
Palantir CEO Alex Karp on why he believes AI labs want to be nationalized:
"First you migrate all the IP to your business, and then because you need capped liability, both vis-à-vis the business and vis-à-vis the dangers, you've got to migrate the business to the government."
"That is the core philosophy of the EA movement imposed upon a business."
Via @CNBC
10
this essentially proves that a massive human teleop data is not necessary for robotics. once we reach RSI(which we likely have), the model only requires a VLA sandbox to continuously run sim-RL, which enables it to resolve and grasp physical principles so robots can operate in the real world.
GPT-6 Astra scored 95% on a robot control task, up from Fable 5.1's 40%, with 6.2x fewer output tokens at 2.3x lower cost. 🧵
1
52
