Visible device: cuda Seed used: 0 Batch size: 64 Epochs: 40 Learning rate: 1e-05 Entropy weight: 0.01 Regularization weight: 0.0 Only use multiwoz like domains: False We use: 1.0% of the data Dialogue order used: 0 Vectorizer: Data set used is multiwoz21 We filter state by active domains: True Vectorizer: Data set used is multiwoz21 Embedding semantic descriptions: True Embedded descriptions successfully. Size: torch.Size([338, 768]) Data set used for descriptions: multiwoz21 We use Roberta to embed actions. Loaded model from experiments/seed0/save/supervised.pol.mdl Start training Epoch: 0 Average actions: 1.9973957538604736 Average target actions: 2.5520834922790527 Precision: 0.09615384615384616 Recall: 0.07462686567164178 F1: 0.08403361344537816 <> epoch 0: saved network to mdl Best Precision: 0.09615384615384616 Best Recall: 0.07462686567164178 Best F1: 0.08403361344537816 Epoch: 1 Precision: 0.09615384615384616 Recall: 0.07462686567164178 F1: 0.08403361344537816 Best Precision: 0.09615384615384616 Best Recall: 0.07462686567164178 Best F1: 0.08403361344537816 Epoch: 2 Average actions: 2.3515625 Average target actions: 2.6197917461395264 Precision: 0.10526315789473684 Recall: 0.08955223880597014 F1: 0.0967741935483871 <> epoch 2: saved network to mdl Best Precision: 0.10526315789473684 Best Recall: 0.08955223880597014 Best F1: 0.0967741935483871 Epoch: 3 Precision: 0.10526315789473684 Recall: 0.08955223880597014 F1: 0.0967741935483871 Best Precision: 0.10526315789473684 Best Recall: 0.08955223880597014 Best F1: 0.0967741935483871 Epoch: 4 Average actions: 1.6770832538604736 Average target actions: 2.8567709922790527 Precision: 0.1347517730496454 Recall: 0.0945273631840796 F1: 0.11111111111111112 <> epoch 4: saved network to mdl Best Precision: 0.1347517730496454 Best Recall: 0.0945273631840796 Best F1: 0.11111111111111112 Epoch: 5 Precision: 0.1347517730496454 Recall: 0.0945273631840796 F1: 0.11111111111111112 Best Precision: 0.1347517730496454 Best Recall: 0.0945273631840796 Best F1: 0.11111111111111112 Epoch: 6 Average actions: 1.9088542461395264 Average target actions: 2.7213542461395264 Precision: 0.12080536912751678 Recall: 0.08955223880597014 F1: 0.10285714285714286 Best Precision: 0.1347517730496454 Best Recall: 0.0945273631840796 Best F1: 0.11111111111111112 Epoch: 7 Precision: 0.12080536912751678 Recall: 0.08955223880597014 F1: 0.10285714285714286 Best Precision: 0.1347517730496454 Best Recall: 0.0945273631840796 Best F1: 0.11111111111111112 Epoch: 8 Average actions: 2.0572915077209473 Average target actions: 2.8229167461395264 Precision: 0.12903225806451613 Recall: 0.09950248756218906 F1: 0.11235955056179776 <> epoch 8: saved network to mdl Best Precision: 0.1347517730496454 Best Recall: 0.09950248756218906 Best F1: 0.11235955056179776 Epoch: 9 Precision: 0.12903225806451613 Recall: 0.09950248756218906 F1: 0.11235955056179776 Best Precision: 0.1347517730496454 Best Recall: 0.09950248756218906 Best F1: 0.11235955056179776 Epoch: 10 Average actions: 2.0911459922790527 Average target actions: 2.6875 Precision: 0.11612903225806452 Recall: 0.08955223880597014 F1: 0.10112359550561797 Best Precision: 0.1347517730496454 Best Recall: 0.09950248756218906 Best F1: 0.11235955056179776 Epoch: 11 Precision: 0.11612903225806452 Recall: 0.08955223880597014 F1: 0.10112359550561797 Best Precision: 0.1347517730496454 Best Recall: 0.09950248756218906 Best F1: 0.11235955056179776 Epoch: 12 Average actions: 2.0833332538604736 Average target actions: 2.5859375 Precision: 0.11976047904191617 Recall: 0.09950248756218906 F1: 0.10869565217391305 Best Precision: 0.1347517730496454 Best Recall: 0.09950248756218906 Best F1: 0.11235955056179776 Epoch: 13 Precision: 0.11976047904191617 Recall: 0.09950248756218906 F1: 0.10869565217391305 Best Precision: 0.1347517730496454 Best Recall: 0.09950248756218906 Best F1: 0.11235955056179776 Epoch: 14 Average actions: 2.1119790077209473 Average target actions: 2.7213542461395264 Precision: 0.16778523489932887 Recall: 0.12437810945273632 F1: 0.14285714285714285 <> epoch 14: saved network to mdl Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 15 Precision: 0.16778523489932887 Recall: 0.12437810945273632 F1: 0.14285714285714285 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 16 Average actions: 1.7994792461395264 Average target actions: 2.5520834922790527 Precision: 0.10135135135135136 Recall: 0.07462686567164178 F1: 0.08595988538681948 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 17 Precision: 0.10135135135135136 Recall: 0.07462686567164178 F1: 0.08595988538681948 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 18 Average actions: 2.0572915077209473 Average target actions: 2.7552084922790527 Precision: 0.13548387096774195 Recall: 0.1044776119402985 F1: 0.11797752808988765 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 19 Precision: 0.13548387096774195 Recall: 0.1044776119402985 F1: 0.11797752808988765 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 20 Average actions: 1.9661457538604736 Average target actions: 2.7213542461395264 Precision: 0.1118421052631579 Recall: 0.0845771144278607 F1: 0.0963172804532578 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 21 Precision: 0.1118421052631579 Recall: 0.0845771144278607 F1: 0.0963172804532578 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 22 Average actions: 1.9557292461395264 Average target actions: 2.5520834922790527 Precision: 0.07741935483870968 Recall: 0.05970149253731343 F1: 0.06741573033707865 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 23 Precision: 0.07741935483870968 Recall: 0.05970149253731343 F1: 0.06741573033707865 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 24 Average actions: 2.0833334922790527 Average target actions: 2.8229167461395264 Precision: 0.09090909090909091 Recall: 0.06965174129353234 F1: 0.07887323943661972 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 25 Precision: 0.09090909090909091 Recall: 0.06965174129353234 F1: 0.07887323943661972 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 26 Average actions: 1.7135417461395264 Average target actions: 2.6197917461395264 Precision: 0.145985401459854 Recall: 0.09950248756218906 F1: 0.1183431952662722 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 27 Precision: 0.145985401459854 Recall: 0.09950248756218906 F1: 0.1183431952662722 Best Precision: 0.16778523489932887 Best Recall: 0.12437810945273632 Best F1: 0.14285714285714285 Epoch: 28 Average actions: 2.0364584922790527 Average target actions: 2.5520834922790527 Precision: 0.16891891891891891 Recall: 0.12437810945273632 F1: 0.14326647564469916 <> epoch 28: saved network to mdl Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 29 Precision: 0.16891891891891891 Recall: 0.12437810945273632 F1: 0.14326647564469916 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 30 Average actions: 2.0026040077209473 Average target actions: 2.3828125 Precision: 0.16216216216216217 Recall: 0.11940298507462686 F1: 0.13753581661891118 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 31 Precision: 0.16216216216216217 Recall: 0.11940298507462686 F1: 0.13753581661891118 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 32 Average actions: 1.8046875 Average target actions: 2.6875 Precision: 0.12142857142857143 Recall: 0.0845771144278607 F1: 0.09970674486803519 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 33 Precision: 0.12142857142857143 Recall: 0.0845771144278607 F1: 0.09970674486803519 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 34 Average actions: 1.9348957538604736 Average target actions: 2.6875 Precision: 0.12162162162162163 Recall: 0.08955223880597014 F1: 0.10315186246418337 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 35 Precision: 0.12162162162162163 Recall: 0.08955223880597014 F1: 0.10315186246418337 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 36 Average actions: 2.0989584922790527 Average target actions: 2.484375 Precision: 0.14743589743589744 Recall: 0.11442786069651742 F1: 0.1288515406162465 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 37 Precision: 0.14743589743589744 Recall: 0.11442786069651742 F1: 0.1288515406162465 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 38 Average actions: 2.0260415077209473 Average target actions: 2.5520834922790527 Precision: 0.1456953642384106 Recall: 0.10945273631840796 F1: 0.12499999999999997 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916 Epoch: 39 Precision: 0.1456953642384106 Recall: 0.10945273631840796 F1: 0.12499999999999997 Best Precision: 0.16891891891891891 Best Recall: 0.12437810945273632 Best F1: 0.14326647564469916