Nano-vLLM: A lightweight vLLM implementation built from scratchgithub.com5 points·by platers·last year·0 comments
Synchronizing Pong to music with constrained optimizationvictortao.substack.com321 points·by platers·2 years ago·37 comments
platers·7 months ago·discussthis would pair well with https://gonzoml.substack.com/p/neurips-2025-best-papers-in-c...
platers·11 months ago·discussI'm struggling to understand where the gains are coming from. What is the intuition for why DiT training was so inefficient?
platers·11 months ago·discusshttps://suno.com/studio-waitlist Just a waitlist so far, but looks like this is the direction suno is going
platers·last year·discussIt sounds very similar to suno v3.5 (including the audio quality) Likely they trained on suno generations.
platers·last year·discussFlash attention natively supports packing multiple variable length sequences into a single call, what is the advantage of jagged flash attention?
platers·2 years ago·discussIs this wrapping Suno? Not going to make an account with only generic marketing copy and no samples.