fanasina/y_project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

second try, not yet working

Browse Source
This commit is contained in:
fanasina
2024-06-21 11:36:28 +02:00
parent 582d7a6a70
commit 824396f901
1 changed files with 16 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
deepQlearn_0/src/deepQlearning/learn_to_drive.c
+16
View File
@@ -205,6 +205,22 @@ void train_qlearning(struct RL_agent * rlAgent,
}else { }else {
experimental_values->x[action] = car_status->reward + rlAgent->qlearnParams->gamma * MAX_ARRAY_TYPE_FLOAT(next_action_value->x, next_action_value->dim->rank) ; experimental_values->x[action] = car_status->reward + rlAgent->qlearnParams->gamma * MAX_ARRAY_TYPE_FLOAT(next_action_value->x, next_action_value->dim->rank) ;
} }
// ***
neurons_TYPE_FLOAT *tmp=NULL, *ttmp=NULL, *base = net_main;
init_copy_in_out_networks_from_tensors_TYPE_FLOAT(base,base->output , experimental_values );\
tmp=net_main->next_layer;\
while(tmp){\
calc_out_neurons_TYPE_FLOAT(tmp);\
ttmp = tmp;\
tmp = tmp->next_layer;\
}\
while(ttmp != base){\
calc_delta_neurons_TYPE_FLOAT(ttmp);\
update_weight_neurons_TYPE_FLOAT(ttmp);\
ttmp = ttmp->prev_layer;\
}\
// ***
float new_value = ( (net_main->learning_rate < qlParams->minimum_threshold_learning_rate /*0.0001*/) ? net_main->learning_rate :(net_main->learning_rate ) * qlParams->factor_update_learning_rate /*0.995*/ ); float new_value = ( (net_main->learning_rate < qlParams->minimum_threshold_learning_rate /*0.0001*/) ? net_main->learning_rate :(net_main->learning_rate ) * qlParams->factor_update_learning_rate /*0.995*/ );
UPDATE_ATTRIBUTE_NEURONE_IN_ALL_LAYERS(TYPE_FLOAT, net_main, learning_rate, new_value); UPDATE_ATTRIBUTE_NEURONE_IN_ALL_LAYERS(TYPE_FLOAT, net_main, learning_rate, new_value);