ICYMI: OpenClaw got upgraded to support newly released GPT-5.4 and Gemini 3.1 Flash Lite models.
LLMS
-
New Paper Evaluates AI Coding Agents on Long-Horizon Codebase Maintenance
By
–
Even the SoTA models still struggle with long-horizon code maintenance! “SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration” Right now, most coding-agent evals are static, where you solve a single issue, and pass one-off tests. This paper
-
The Chinese Room Revisited for the Modern Era
By
–
The Chinese Room thought experiment runs like this, updated for modern times. A man who speaks no Chinese is locked in a room. He receives a card bearing a Chinese character. The man looks up the character in a table, and retrieves 16,384 numbers, each recorded to 3
-
Microsoft Launches Free Generative AI Course in Spanish
By
–
BREAKING NEWS: Microsoft has officially released a free Generative Artificial Intelligence course, available in Spanish. The course includes:
• 21 step-by-step lessons
• Code examples and explanatory videos
• Coverage from basic concepts to building AI agents and training models I'll share the link— -
Top AI Papers of the Week: Agents, LLMs and Theory
By
–
The Top AI Papers of the Week (March 1 – March 8) – NeuroSkill
– ParamMem
– Numina-Lean-Agent
– Bayesian Teaching for LLMs
– Auton Agentic AI Framework
– Theory of Mind in Multi-Agent LLMs
– Why LLMs Form Geometric Representations Read on for more: -
Mistral AI Adds Workflows to Le Chat
By
–
Mistral AI is working on adding Workflows support to Le Chat. Workflows have been in development on Mistral Playground since last year and appear to be nearing a broader release.
-
Amazon’s Rufus AI generates free Claude tokens
By
–
LMAO! It works. Amazon's Rufus AI as a free Claude token generator.
-
Karpathy Open-Sources Autoresearch: An AI Agent for ML Experiments
By
–
Karpathy just open-sourced autoresearch. It runs 100 ML experiments overnight on a single GPU. The agent writes the code, runs the training, iterates and keeps what works. Your entire job is maintaining a single Markdown file, one that describes the research strategy. What to
-
GPT-5.4 Pro for Advanced Physics Research Problems
By
–
gpt-5.4 pro for research-level physics problems:
-
Technical Analysis of FlashAttention-4 Performance Bottlenecks
By
–
"FlashAttention-4" This new iteration of Flash Attention shows that on NVIDIA Blackwell GPUs the new bottleneck in Transformer attention isn’t matmul anymore but softmax + shared-memory traffic. So the latest designs stop treating attention as just GEMMs and start co-designing