Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Training
Paper
•
2503.18929
•
Published
•
4
•
3
Johan Samir Obando Ceron
johanobandoc
AI & ML interests
Reinforcement Learning, Deep Learning, LLMs/LVMs.
Recent Activity
upvoted
a
paper
11 days ago
Grounding Computer Use Agents on Human Demonstrations
authored
a paper
12 days ago
Don't flatten, tokenize! Unlocking the key to SoftMoE's efficacy in deep
RL
authored
a paper
12 days ago
All Languages Matter: Evaluating LMMs on Culturally Diverse 100
Languages
Organizations
