Inference Optimization for MiniMax Sparse Attentiontogether.ai1 points·by zagwdt·в прошлом месяце·0 comments
DeepSeek V4 in vLLM: Efficient Long-Context Attentionvllm-website-pdzeaspbm-inferact-inc.vercel.app3 points·by zagwdt·3 месяца назад·0 comments
Introspective Diffusion Language Modelsintrospective-diffusion.github.io281 points·by zagwdt·3 месяца назад·55 comments
EinsteinArena: Harnessing the collective intelligence of agents in the wildeinsteinarena.com5 points·by zagwdt·3 месяца назад·0 comments
TorchSpec: Speculative Decoding Training at Scalepytorch.org2 points·by zagwdt·4 месяца назад·0 comments
CoderForge-Preview: SOTA open dataset for training efficient coding agentstogether.ai1 points·by zagwdt·5 месяцев назад·0 comments
Two years of vector search at Notion: 10x scale, 1/10th costnotion.com2 points·by zagwdt·5 месяцев назад·0 comments
Consistency diffusion language models: Up to 14x faster, no quality losstogether.ai219 points·by zagwdt·5 месяцев назад·96 comments
zagwdt·3 месяца назад·discussngl lost alot of trust in cc after reading this, specially point 1how do you just do that to millions of users building prod code with your shit
how do you just do that to millions of users building prod code with your shit