Nano-vLLM: A lightweight vLLM implementation built from scratchgithub.com5 ポイント·投稿者 platers·昨年·0 コメント
Synchronizing Pong to music with constrained optimizationvictortao.substack.com321 ポイント·投稿者 platers·2 年前·37 コメント
platers·8 か月前·議論this would pair well with https://gonzoml.substack.com/p/neurips-2025-best-papers-in-c...
platers·11 か月前·議論I'm struggling to understand where the gains are coming from. What is the intuition for why DiT training was so inefficient?
platers·11 か月前·議論https://suno.com/studio-waitlist Just a waitlist so far, but looks like this is the direction suno is going
platers·昨年·議論It sounds very similar to suno v3.5 (including the audio quality) Likely they trained on suno generations.
platers·昨年·議論Flash attention natively supports packing multiple variable length sequences into a single call, what is the advantage of jagged flash attention?
platers·2 年前·議論Is this wrapping Suno? Not going to make an account with only generic marketing copy and no samples.