Andrej Karpathy builds video platform with Claude AI
He distilled 43 million Hacker News comments into 6,454 videos using Claude Opus 5.5.
Stories per day · 30 days
Relatedclustering1computer-use1cryptanalysis1defense1extraction1linear-attention1llama.cpp1microsoft1nvidia1
8 stories, newest first · one card per story
Stories tagged #model
Guide to running NVIDIA Dory 4B with llama.cpp
Loop-by-loop instructions for deploying the 4B parameter NVIDIA Dory model using llama.cpp.
Muir Model Improves Linear Attention for Large Contexts
Muir outperforms transformers and gated delta models on long-context tasks while training faster
Adaptive clustering breaks TraintoDefend defense in model extraction
SimBreak raises signature recovery from 0% to 100% in protected networks.
Microsoft releases Mage-VL-Cowork 4B parameter model
Mage-VL-Cowork is a 4B parameter hybrid computer use model achieving 66.0% on OSWorld-Verified.
AI performance costs are falling faster than those of any previous technology
AI performance costs are falling faster than any previous technology, with 47% quarterly and 13x annual drops in average costs.
Rare Event Estimation via Iterative Unalignment
Research paper on estimating rare events from agent outputs, proposing a new importance sampling method with large efficiency gains.
Fastly gives enterprises real-time control over AI models and agents
Fastly launches AI Runtime Control, AI Firewall, and API Security capabilities to give enterprises real-time control over AI models, applications, and agents.