Stories by limoce

FlexAttention: The Flexibility of PyTorch with the Performance of FlashAttention

MiniCPM-v2.6: GPT-4V Level MLLM for Single/Multi Image and Video on Your Phone

MindSearch: LLM-Based Web Search Engine Similar to Perplexity.ai and SearchGPT

Turbo Sparse: Achieving LLM SOTA Performance with Minimal Activated Parameters

PowerInfer-2: Fast Large Language Model Inference on a Smartphone

Large-scale photonic chiplet Taichi empowers 160TOPS/W AI

Asterinas: OS kernel written in Rust and providing Linux-compatible ABI

Mq-deadline scalability improvements (with more than 100% improvement)

Researchers Create First Functional Semiconductor Made from Graphene

Wayland Enjoyed Many Successes in 2023

Improving our safety with a physical quantities and units library

Nesting chinstrap penguins sleep by seconds-long microsleeps

PowerInfer: High-Speed Large Language Model Serving on Consumer-Grade GPUs

Zpoline: System Call Hook for Linux

Randomized Single-Source Shortest Path Algo. On Undirected Real-Weighted Graphs

ChatGPT powered Rust proc macro that generates code at compile-time

Memcpy is faster than memset on Intel i7 12700 with glibc 2.36

Year-in-search-trends: Visualization of search interest over time

Make call_rcu() lazy to save power

IOMMUFD

1
2
3
4
5
6