Show HN: We quantized Qwen3.6-35B-A3B to 2-bit: 12.3 GB, 225 tok/s on a 4090huggingface.co 3yzh1mo3 comments
MyAI101: Foundational AI literacy for students, teachers and curious adultsmyai101.com 5yzh1y3 comments
Speedup Megatron-LM from 10% to 30% with Zero Bubble Pipeline Parallelismgithub.com/sail-sg 4yzh2y2 comments
HloEnv: An environment based on XLA for DL compiler optimization researchgithub.com/sail-sg 2yzh3ydiscuss
AI Research Trends:Sort ArXiv papers by its popularity on social media sitesmavenlin.github.io 52yzh4y10 comments
Jittor: A Just-in-time(JIT) deep learning framework from Tsinghua Universitygithub.com/Jittor 2yzh6ydiscuss
Tencent open sourced its distributed graph computing framework: Platogithub.com/Tencent 49yzh6y1 comment