#optimization
Every summary, chronological. Filter by category, tag, or source from the rail.
Optimizing Transformer Inference with FlashNorm
FlashNorm accelerates transformer inference by folding RMS norm gains into projection weights and parallelizing normalization and matrix multiplication via custom CUDA kernels.
AI EngineerAutomating QUBO Formulation from Natural Language
This paper introduces a method to bridge the gap between human-readable optimization problem descriptions and the mathematical rigor of Quadratic Unconstrained Binary Optimization (QUBO) using LLMs.
Decoupling RL Rollout Fleets from Training Clusters via Stitch
By exploiting the fact that Adam-optimized model updates are sparse in low-precision serving views, you can sync rollout weights via 500MB patches instead of 500GB checkpoints, enabling global, elastic RL training.
AI EngineerCompression at the Edge: Strategies for Efficient AI
Compression is not just about fitting models on consumer hardware; it is a strategic necessity for democratizing intelligence, increasing concurrency, and reducing operational costs by leveraging selective quantization and architecture-aware optimization.
AI EngineerAgentic Aggregators for Electric Bus Fleet Management
Agentic systems can optimize electric bus fleets by balancing grid flexibility and operational constraints, but profit-oriented configurations risk extracting value from public transport operators.
Flash-KMeans: Accelerating Exact Clustering on GPUs
Flash-KMeans optimizes Lloyd's k-means algorithm for GPUs by restructuring dataflow to eliminate HBM bottlenecks, achieving up to 200x speedups over FAISS without sacrificing mathematical accuracy.
COAgents: A Multi-Agent Framework for Routing Optimization
COAgents is a multi-agent framework designed to navigate complex search spaces in routing problems by combining collaborative agent intelligence with optimization techniques.
How Adam's Variance Normalization Fixes SGD's Frequency Bias
Standard SGD fails to optimize rare tokens because they receive infrequent gradient updates. Adam solves this by using variance normalization to automatically amplify the effective learning rate for rare parameters.
Showing 8 of 8