You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Two partially independent ideas on alternative architecture of learning system. In order to increase performance and scalability.
Distributed system. Training on central server (all steps for all maps required). Validation (SVM running) can de done independently on several machines. Each machine plays part of maps. In combination with second idea it may be possible to send maps by chunks for better load balancing.
System without communication on each step. New learning cycle:
train (requires all steps for all maps collected)
Export model to ONNX
Send new model to each SVM
Each SVM plays maps and dumps steps for them.
SVM checks whether steps should be updated or merged and if so, send them to the agent.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Two partially independent ideas on alternative architecture of learning system. In order to increase performance and scalability.
Beta Was this translation helpful? Give feedback.
All reactions