v1.2.0

stable training

v1.1.0

Stable training :

• batch training
• some fix
• new CLI draft

v1.0.0

• model trains well
• fix objective function
• fix truncated normal
• add trained model state dict
• updated spider json
• some update / fix