How can I run the DAMD with the L_sto loss ?

[FlyingCat-fa]

Thanks for your nice work!

I try run the damd model. After Estimated Behavior Policy, I only get the fn_Gs_10_0.0_act_soft.json, but not fn_Qs_10_0.0_act_soft.json. Thus, when I run the end-2-end damd, I only can use the L_det loss, but can not use the L_sto loss. What's the problem ? How can I get the fn_Qs_10_0.0_act_soft.json ?