» Tag
machine-learning
163 postsHollow-LLM Attack: Ghost Weights That Fool Zero-Knowledge LLM Verification
The Hollow-LLM attack examines the impact of ghost weights on zero-knowledge verification.
Kimi Delta Attention: A New Linear Attention Approach
Kimi Delta Attention is the latest innovation in linear attention mechanisms. Explore its important mathematical foundations for engineers.
Model Routing: Simple in Theory, Complex in Practice
Model routing requires system optimization to balance cost and performance effectively.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comWe Built a Day-0 API for Kimi K3
Baseten has introduced day-0 support for Kimi K3 on its Model APIs, addressing the challenges of its 2.8T parameters.
My hobby compiler now compiles a company's production app
Edge Python allows users to run Python business logic on the client side, offering new possibilities with a fast and lightweight compiler.
LoongForge: High-Performance Training System for Embodied Models
LoongForge-Embodied offers a PyTorch-based training subsystem for VLA and WAM models.
Echo: Fable-level Results Using Open-weight Models at a Third of the Cost
Echo delivers Fable-level results using open-weight models while reducing costs.
Reducing Token Usage Through Representation Engineering
Insights on reducing token usage with the Caveman package and the role of activation steering.
GEMM Performance Measurement Methodology Guidelines
Develop reliable and reproducible methods for measuring GEMM performance.
Understanding Why AI Agents Get Stuck in Loops
Explore the challenges AI agents face in task completion and strategies to overcome them.