Upload folder using huggingface_hub

.ipynb_checkpoints/log-checkpoint.txt

@@ -0,0 +1,9 @@
+{"train_lr": 9.999999999999999e-05, "train_min_lr": 5.000000000000001e-06, "train_loss": 0.6890836213086102, "train_loss_scale": 24256.192192192193, "train_weight_decay": 0.009999999999999919, "train_grad_norm": Infinity, "val_score": 58.749999046325684, "epoch": 0, "n_parameters": 223636442}
+{"train_lr": 0.00019771485864187942, "train_min_lr": 9.885742932093977e-06, "train_loss": 0.7736698335355466, "train_loss_scale": 16384.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 24.647449934446776, "val_score": 58.45703089237213, "epoch": 1, "n_parameters": 223636442}
+{"train_lr": 0.00018438163965014831, "train_min_lr": 9.219081982507408e-06, "train_loss": 0.6634077656197477, "train_loss_scale": 11611.483483483484, "train_weight_decay": 0.009999999999999919, "train_grad_norm": NaN, "val_score": 57.28515565395355, "epoch": 2, "n_parameters": 223636442}
+{"train_lr": 0.00015972449029292443, "train_min_lr": 7.986224514646211e-06, "train_loss": 0.5795572266295865, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 23.064132601649195, "val_score": 58.43749928474426, "epoch": 3, "n_parameters": 223636442}
+{"train_lr": 0.0001274972380422297, "train_min_lr": 6.374861902111488e-06, "train_loss": 0.498919264869289, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 21.084254748828418, "val_score": 59.23828089237213, "epoch": 4, "n_parameters": 223636442}
+{"train_lr": 9.260618991246354e-05, "train_min_lr": 4.630309495623178e-06, "train_loss": 0.42811756480384516, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 19.145860155065495, "val_score": 58.789061188697815, "epoch": 5, "n_parameters": 223636442}
+{"train_lr": 6.036319169324366e-05, "train_min_lr": 3.018159584662179e-06, "train_loss": 0.36774679338251864, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 17.216959914645635, "val_score": 59.667967796325684, "epoch": 6, "n_parameters": 223636442}
+{"train_lr": 3.5676947579939585e-05, "train_min_lr": 1.7838473789969771e-06, "train_loss": 0.32623576064099064, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 16.371801275390762, "val_score": 60.13671803474426, "epoch": 7, "n_parameters": 223636442}
+{"train_lr": 2.2305714457441824e-05, "train_min_lr": 1.115285722872091e-06, "train_loss": 0.2990256009718498, "train_loss_scale": 12915.315315315316, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 15.5836338116242, "val_score": 59.257811546325684, "epoch": 8, "n_parameters": 223636442}

checkpoint-best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70b54489ed2d74daa71fb27d2235491204fecc73db25d13e941f632ba5736708
+size 2684267121

log.txt

@@ -0,0 +1,9 @@
+{"train_lr": 9.999999999999999e-05, "train_min_lr": 5.000000000000001e-06, "train_loss": 0.6890836213086102, "train_loss_scale": 24256.192192192193, "train_weight_decay": 0.009999999999999919, "train_grad_norm": Infinity, "val_score": 58.749999046325684, "epoch": 0, "n_parameters": 223636442}
+{"train_lr": 0.00019771485864187942, "train_min_lr": 9.885742932093977e-06, "train_loss": 0.7736698335355466, "train_loss_scale": 16384.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 24.647449934446776, "val_score": 58.45703089237213, "epoch": 1, "n_parameters": 223636442}
+{"train_lr": 0.00018438163965014831, "train_min_lr": 9.219081982507408e-06, "train_loss": 0.6634077656197477, "train_loss_scale": 11611.483483483484, "train_weight_decay": 0.009999999999999919, "train_grad_norm": NaN, "val_score": 57.28515565395355, "epoch": 2, "n_parameters": 223636442}
+{"train_lr": 0.00015972449029292443, "train_min_lr": 7.986224514646211e-06, "train_loss": 0.5795572266295865, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 23.064132601649195, "val_score": 58.43749928474426, "epoch": 3, "n_parameters": 223636442}
+{"train_lr": 0.0001274972380422297, "train_min_lr": 6.374861902111488e-06, "train_loss": 0.498919264869289, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 21.084254748828418, "val_score": 59.23828089237213, "epoch": 4, "n_parameters": 223636442}
+{"train_lr": 9.260618991246354e-05, "train_min_lr": 4.630309495623178e-06, "train_loss": 0.42811756480384516, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 19.145860155065495, "val_score": 58.789061188697815, "epoch": 5, "n_parameters": 223636442}
+{"train_lr": 6.036319169324366e-05, "train_min_lr": 3.018159584662179e-06, "train_loss": 0.36774679338251864, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 17.216959914645635, "val_score": 59.667967796325684, "epoch": 6, "n_parameters": 223636442}
+{"train_lr": 3.5676947579939585e-05, "train_min_lr": 1.7838473789969771e-06, "train_loss": 0.32623576064099064, "train_loss_scale": 8192.0, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 16.371801275390762, "val_score": 60.13671803474426, "epoch": 7, "n_parameters": 223636442}
+{"train_lr": 2.2305714457441824e-05, "train_min_lr": 1.115285722872091e-06, "train_loss": 0.2990256009718498, "train_loss_scale": 12915.315315315316, "train_weight_decay": 0.009999999999999919, "train_grad_norm": 15.5836338116242, "val_score": 59.257811546325684, "epoch": 8, "n_parameters": 223636442}