An attention head is a differentiable multiplexer.
LLMS
-

LLM Sycophancy and Chatbot-Associated Delusions: A Mind-Blowing 37%
By
–
Holy crap. I knew about sycophancy. But the 37% number below blows my mind. This from an analysis of chat logs in people who experienced chatbot-associated delusions. In over a third of the messages to those users, the LLMs told the users they had (eg)
-

AI Agents Could Push College Graduate Unemployment Above 30%
By
–
#AIAgents could easily send college grad unemployment over 30%, ServiceNow CEO says
by @samantha_subin @cnbc Learn more: https://
bit.ly/3NfapcV #LLM #GenerativeAI #ArtificialIntelligence #MachineLearning -

ODSC AI East 2026: Production AI Engineering Practices and Techniques
By
–
At ODSC AI East 2026, the focus is learning the Engineering Practices behind Production AI systems: Agentic AI, RAG, LLMOps, LLMs, GenAI, Model Fine-Tuning, … Boston + Virtual on April 28–30 Register here: https://
hubs.li/Q04746MR0 via @_odsc -
Quantization Impact on Model Quality and Expert Reduction
By
–
How confident are you with respect to the output quality given the 2-bit quantization and reducing experts from 10 to 4? Did you have a mechanism for measuring that?
-

Running 397B MoE Model on M3 Mac with Efficient Weight Streaming
By
–
Dan says he's got Qwen 3.5 397B-A17B – a 209GB on disk MoE model – running on an M3 Mac at ~5.7 tokens per second using only 5.5 GB of active memory (!) by quantizing and then streaming weights from SSD (at ~17GB/s), since MoE models only use a small subset of their weights for
-

MiniMax-M2.7 model hits SWE-Pro SOTA at 56.22%
By
–

MiniMax-M2.7 is here. → matching Sonnet 4.6 as an agent
→ recovering live incidents in just 3 minutes
→ hitting SOTA in SWE-Pro (56.22%) It even edits your Office files -
Claude and Codex: AI Models on the Back Burner
By
–
claude and codex are now chefs of my back burner
-
Which LLM excels at persuasive copywriting?
By
–
Let's settle this. if we're talking purely about persuasive copywriting quality, which LLM takes the crown?