Shoot those mothers down, PPO!
Shoot those mothers down, PPO!
Is PPO enough for much more interesting tasks than we expected? Abundantly yes.
Is PPO enough for much more interesting tasks than we expected? Abundantly yes.
T-1. 15-16 Ducks - 18 (39 PPO)
T-1. 14-15 Penguins - 18 (42 PPO)
T-3. 24-25 Wings - 17 (34 PPO)*
T-3. 23-24 Canes - 17 (33 PPO)
T-3. 23-24 Devils - 17 (36 PPO)
T-3. 22-23 Oilers - 17 (37 PPO)
T-1. 15-16 Ducks - 18 (39 PPO)
T-1. 14-15 Penguins - 18 (42 PPO)
T-3. 24-25 Wings - 17 (34 PPO)*
T-3. 23-24 Canes - 17 (33 PPO)
T-3. 23-24 Devils - 17 (36 PPO)
T-3. 22-23 Oilers - 17 (37 PPO)
▶️ www.mdpi.com/1422-0067/27...
#PlantScience
▶️ www.mdpi.com/1422-0067/27...
#PlantScience
I think this is a bad system actually.
I think this is a bad system actually.
me: i doubt either of these are covered by my insurance
I do not take well to trainees pushing me on things they don’t know anything about :3
Having a medical background is different than having an insurance background shut the FAWK up 💅
Oh yeah and of COURSE it was a man 👹
www.theatlantic.com/ideas/archiv...
Debugging why PPO SB3 (PyTorch) was working where PPO SBX (Jax) was failing.
Just one line of code different... 🙈 (the initialization of the actor network output)
Debugging why PPO SB3 (PyTorch) was working where PPO SBX (Jax) was failing.
Just one line of code different... 🙈 (the initialization of the actor network output)
So, study why PPO exists (lots of docs / writing on that) and understand that value functions are tricky with LLMs.
Left ppo, right grpo
So, study why PPO exists (lots of docs / writing on that) and understand that value functions are tricky with LLMs.
Left ppo, right grpo
He harassed me very aggressively making trans phobic remarks and making threatening gestures, I made a police report. months pass