GPT-OSS-Safeguard from @OpenAI is here. Open-weight, safety-tuned, transparent reasoning. Now available in private preview at Cerebras speeds https://
cerebras.ai/build-with-us
RESEARCH
-

OpenAI Releases GPT-OSS-Safeguard Open-Weight Model
By
–
-
Visualizing Llama 3.1 Tensor Operations Through GGML Debug Output
By
–
The three.js visualization parses the output of llama-cpp's ggml debug output (of unsloth llama 3.1) to directly obtain all the tensor calculations happening under the hood. Operations (MUL_MAT, ROPE, RESHAPE, ADD) are grouped into query, key, value, MLP, and residual stream
-
3D Illustrated Transformer: Interactive LLaMA Learning Tool
By
–
Introducing the Illustrated Transformer in 3D 🚀
— alphaXiv (@askalphaxiv) 29 octobre 2025
Fly through LLaMA like never before. See every tensor and operation in motion.
Click any component to reveal the exact lines of code that run it.
A new way to learn and teach LLMs. Try it out in the link below 👇 pic.twitter.com/pBpEZyse2vIntroducing the Illustrated Transformer in 3D Fly through LLaMA like never before. See every tensor and operation in motion. Click any component to reveal the exact lines of code that run it. A new way to learn and teach LLMs. Try it out in the link below
-
Building Conscious AI: Beyond Mimicry to Real Consciousness
By
–
I wrote about the quest to build conscious AI for @WIRED
's AI special issue. Most experts think today's models just mimic human consciousness but some are trying to build models that actually are. -

Compute Trade-offs: Factual Knowledge vs Dense Core Efficiency
By
–
There are two kinds of people in AI: those who are happily surprised a model would know random facts like these (worth burning compute and parameters), and those who think it's a complete waste of both for a dense core.
-
Jeff Dean Praises Team Pushing Machine Learning Boundaries
By
–
I love working with members of this team, who are pushing the boundaries of machine learning!
-
DeepSeek-OCR OmniDocBench: Document Recognition Benchmark
By
–
Detailed report + code: https://
github.com/alphaXiv/DeepS
eek-OCR-OmniDocBench/blob/main/REPORT.md
… Datasets page: http://
alphaxiv.org/datasets/shang
hai-ai-laboratory/omnidocbench
… -

Meta cuts 600 AI jobs, survivors reapply in major restructuring
By
–
Big shakeup at Meta: ~600 AI roles cut across FAIR, product & infra. TBD Lab – the high-stakes superintelligence arm, spared. Wang’s memo: smaller teams = faster decisions, more impact per person. But here’s the twist: those laid off are being told to reapply
-
Thermodynamic Computing: Revolutionary New Approach to Computation
By
–
Excited to see. If you have been tracking these builders you probably are too!
— Bob Gourley – e/acc (@bobgourley) 28 octobre 2025
This is a completely new way of doing compute. Not digital, not quantum, but thermodynamic. https://t.co/I84j4PyzH2Excited to see. If you have been tracking these builders you probably are too! This is a completely new way of doing compute. Not digital, not quantum, but thermodynamic.
-
Measuring AI Progress: Velocity and Acceleration Matter Most
By
–
In order to track the pace of change (and promise) in AI, you have to be looking at velocity and acceleration. There is a very big difference between "AI scored 80% of this one benchmark" and "AI scored 80% on this one benchmark and last month it only scored 10%". You need to