开启时会模糊预览图，关闭后正常显示
Booting up the livestream now. Will be fighting the last couple numerical mismatches in PufferLib's new CUDA kernels. Fully fused loss fn, gating, scan, etc.
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.
60 set Sunday complete. 3x ~8-12 reps on: Chest (9 sets): Incline smith bench, cable crossover, machine press Shoulders (12 sets): Machine press, cable raise, dumbbell press, face pull Back (12 sets): Pullups, smith rows, lat pulldown, low row Arms (12 sets): Dumbbell preacher, cable curl, cable kickback, dumbbell skull crusher Legs (6 sets.. I run a lot okay): Extension, lying curl Abs (9 sets): Dragon flag, cable twist, cable crunch
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.
Today, I wrote three kernels fusing various gates and scans in our recurrent cell and a fused PPO loss kernel. Result is +2M steps/second training before I even start optimizing. Streamed >12 hours of dev today. Tomorrow, I lift weights all day and relax. Star pufferlib. Gn
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.
Live RL Research on PufferLib w/ Joseph Suarez
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.
Live RL Research on PufferLib w/ Joseph Suarez
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.
PufferLib is leading small-model RL, all OSS! The government isn't feeding the puffer any gold bars, but you can feed him a gold star on github
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.