Zero-Copy GPU Inference from WebAssembly on Apple Silicon(abacusnoir.com)

103 points by agambrahma 6 months ago | 38 comments

NixOS and reproducible builds could have detected the xz backdoor(luj.fr)

260 points by birdculture 1 year ago | 112 comments

What Is RLCD? The Secret Behind Jev(di-zhang-llm.github.io)

51 points by tnspacetime 1 month ago | 6 comments

Blog post by Fudan PhD candidate Di Zhang explaining RLCD, a calibrated reward modeling approach building on PPRM and Plackett-Luce, and its role in Jev.

How GLM built its own inference infrastructure(z.ai)

373 points by whiteros_e 1 month ago | 261 comments

Z.ai research blog post describing how its GLM model was used to build its own inference infrastructure, framed as a step toward recursive self-improvement.

Detecting LLM-Generated Texts with “Classical” Machine Learning(blog.lyc8503.net)

219 points by uneven9434 3 months ago | 163 comments

AI 2027(ai-2027.com)

789 points by Tenoke 1 year ago | 518 comments

AI 2027 is a scenario by Daniel Kokotajlo, Scott Alexander and others forecasting the impact of superhuman AI over the next decade, with interactive capability charts.