20241026-19:07:25 argv ./main.py --nb_models=1 --gpus=0 --model=352M --batch_size=5 --eval_batch_size=5 --result_dir=results_culture_352M --learning_rate=2e-4
20241026-19:07:25 args.log_filename train.log
20241026-19:07:25 args.result_dir results_culture_352M
20241026-19:07:25 args.seed 0
20241026-19:07:25 args.resume False
20241026-19:07:25 args.optimizer adam
20241026-19:07:25 args.nb_warmup_samples 5000
20241026-19:07:25 args.nb_epochs 10000
20241026-19:07:25 args.batch_size 5
20241026-19:07:25 args.train_batch_size None
20241026-19:07:25 args.eval_batch_size 5
20241026-19:07:25 args.nb_train_samples 50000
20241026-19:07:25 args.nb_test_samples 2500
20241026-19:07:25 args.nb_c_quizzes 5000
20241026-19:07:25 args.c_quiz_multiplier 1
20241026-19:07:25 args.learning_rate 0.0002
20241026-19:07:25 args.gradient_clipping None
20241026-19:07:25 args.nb_have_to_be_correct 3
20241026-19:07:25 args.nb_have_to_be_wrong 1
20241026-19:07:25 args.nb_mistakes_to_be_wrong 5
20241026-19:07:25 args.model_type standard
20241026-19:07:25 args.model 352M
20241026-19:07:25 args.dim_model 1024
20241026-19:07:25 args.dim_keys 64
20241026-19:07:25 args.dim_hidden 2048
20241026-19:07:25 args.nb_heads 8
20241026-19:07:25 args.nb_blocks 48
20241026-19:07:25 args.dropout 0.5
20241026-19:07:25 args.nb_threads 1
20241026-19:07:25 args.gpus 0
20241026-19:07:25 args.nb_models 1
20241026-19:07:25 args.diffusion_nb_iterations 25
20241026-19:07:25 args.diffusion_proba_corruption 0.05
20241026-19:07:25 args.accuracy_to_make_c_quizzes 0.95
20241026-19:07:25 args.proba_prompt_noise 0.05
20241026-19:07:25 args.proba_hint 0.25
20241026-19:07:25 args.quizzes None
20241026-19:07:25 args.test None
20241026-19:07:25 args.grids_world_tasks replace_color,translate,grow,frame
20241026-19:07:39 main_device cuda:0 gpus ['cuda:0']
20241026-19:07:39 vocabulary_size 11
20241026-19:07:40 nb_parameters 352699403 (352M)
20241026-19:07:40 wrote state.pth
20241026-19:07:40 --- epoch 0 ----------------------------------------
20241026-19:07:40 current_test_accuracies 0.0000
20241026-19:07:40 no_c_quiz
20241026-19:07:40 weakest_accuracies [0.0]
20241026-19:07:53 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-19:24:50 train_loss 0 model 0 0.545583251953125
20241026-19:24:50 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-19:25:06 test_loss 0 model 0 0.119640625
20241026-19:25:06 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-19:25:24 test_accuracy 0 model 0 nb_correct 56 / 2500 (2.24%)
20241026-19:25:24 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-19:25:55 wrote ae_*.pth
20241026-19:25:55 epoch_duration 18min14s next_finish 19:44:10
20241026-19:25:55 wrote state.pth
20241026-19:25:55 --- epoch 1 ----------------------------------------
20241026-19:25:55 current_test_accuracies 0.0224
20241026-19:25:55 no_c_quiz
20241026-19:25:55 weakest_accuracies [0.0224]
20241026-19:25:55 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-19:44:19 train_loss 1 model 0 0.11347545166015625
20241026-19:44:19 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-19:44:35 test_loss 1 model 0 0.07401565551757812
20241026-19:44:35 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-19:44:51 test_accuracy 1 model 0 nb_correct 637 / 2500 (25.48%)
20241026-19:44:51 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-19:45:21 wrote ae_*.pth
20241026-19:45:21 epoch_duration 19min26s next_finish 20:04:48
20241026-19:45:21 wrote state.pth
20241026-19:45:21 --- epoch 2 ----------------------------------------
20241026-19:45:21 current_test_accuracies 0.2548
20241026-19:45:21 no_c_quiz
20241026-19:45:21 weakest_accuracies [0.2548]
20241026-19:45:22 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-20:03:47 train_loss 2 model 0 0.0843360122680664
20241026-20:03:47 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-20:04:03 test_loss 2 model 0 0.05526161193847656
20241026-20:04:03 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-20:04:19 test_accuracy 2 model 0 nb_correct 929 / 2500 (37.16%)
20241026-20:04:19 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-20:04:49 wrote ae_*.pth
20241026-20:04:49 epoch_duration 19min27s next_finish 20:24:17
20241026-20:04:49 wrote state.pth
20241026-20:04:49 --- epoch 3 ----------------------------------------
20241026-20:04:49 current_test_accuracies 0.3716
20241026-20:04:49 no_c_quiz
20241026-20:04:49 weakest_accuracies [0.3716]
20241026-20:04:49 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-20:23:17 train_loss 3 model 0 0.07174256896972656
20241026-20:23:17 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-20:23:34 test_loss 3 model 0 0.048340813636779785
20241026-20:23:34 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-20:23:50 test_accuracy 3 model 0 nb_correct 1225 / 2500 (49.00%)
20241026-20:23:50 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-20:24:20 wrote ae_*.pth
20241026-20:24:20 epoch_duration 19min30s next_finish 20:43:50
20241026-20:24:20 wrote state.pth
20241026-20:24:20 --- epoch 4 ----------------------------------------
20241026-20:24:20 current_test_accuracies 0.4900
20241026-20:24:20 no_c_quiz
20241026-20:24:20 weakest_accuracies [0.49]
20241026-20:24:20 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-20:43:00 train_loss 4 model 0 0.0622249698638916
20241026-20:43:00 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-20:43:15 test_loss 4 model 0 0.0432762336730957
20241026-20:43:15 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-20:43:32 test_accuracy 4 model 0 nb_correct 1427 / 2500 (57.08%)
20241026-20:43:32 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-20:44:02 wrote ae_*.pth
20241026-20:44:02 epoch_duration 19min42s next_finish 21:03:45
20241026-20:44:02 wrote state.pth
20241026-20:44:02 --- epoch 5 ----------------------------------------
20241026-20:44:02 current_test_accuracies 0.5708
20241026-20:44:02 no_c_quiz
20241026-20:44:02 weakest_accuracies [0.5708]
20241026-20:44:02 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-21:02:25 train_loss 5 model 0 0.0573727952003479
20241026-21:02:25 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-21:02:42 test_loss 5 model 0 0.03737013578414917
20241026-21:02:42 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-21:02:59 test_accuracy 5 model 0 nb_correct 1456 / 2500 (58.24%)
20241026-21:02:59 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-21:03:29 wrote ae_*.pth
20241026-21:03:29 epoch_duration 19min26s next_finish 21:22:55
20241026-21:03:29 wrote state.pth
20241026-21:03:29 --- epoch 6 ----------------------------------------
20241026-21:03:29 current_test_accuracies 0.5824
20241026-21:03:29 no_c_quiz
20241026-21:03:29 weakest_accuracies [0.5824]
20241026-21:03:29 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-21:21:55 train_loss 6 model 0 0.052121820497512815
20241026-21:21:55 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-21:22:11 test_loss 6 model 0 0.038954888343811035
20241026-21:22:11 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-21:22:27 test_accuracy 6 model 0 nb_correct 1540 / 2500 (61.60%)
20241026-21:22:27 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-21:22:56 wrote ae_*.pth
20241026-21:22:56 epoch_duration 19min26s next_finish 21:42:22
20241026-21:22:56 wrote state.pth
20241026-21:22:56 --- epoch 7 ----------------------------------------
20241026-21:22:56 current_test_accuracies 0.6160
20241026-21:22:56 no_c_quiz
20241026-21:22:56 weakest_accuracies [0.616]
20241026-21:22:56 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-21:41:19 train_loss 7 model 0 0.04776791572570801
20241026-21:41:19 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-21:41:35 test_loss 7 model 0 0.03266408491134644
20241026-21:41:35 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-21:41:51 test_accuracy 7 model 0 nb_correct 1716 / 2500 (68.64%)
20241026-21:41:51 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-21:42:21 wrote ae_*.pth
20241026-21:42:21 epoch_duration 19min25s next_finish 22:01:46
20241026-21:42:21 wrote state.pth
20241026-21:42:21 --- epoch 8 ----------------------------------------
20241026-21:42:21 current_test_accuracies 0.6864
20241026-21:42:21 no_c_quiz
20241026-21:42:21 weakest_accuracies [0.6864]
20241026-21:42:21 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-22:00:45 train_loss 8 model 0 0.04374433674812317
20241026-22:00:45 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:01:01 test_loss 8 model 0 0.029519215166568757
20241026-22:01:01 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:01:17 test_accuracy 8 model 0 nb_correct 1944 / 2500 (77.76%)
20241026-22:01:17 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-22:01:46 wrote ae_*.pth
20241026-22:01:46 epoch_duration 19min25s next_finish 22:21:11
20241026-22:01:46 wrote state.pth
20241026-22:01:46 --- epoch 9 ----------------------------------------
20241026-22:01:46 current_test_accuracies 0.7776
20241026-22:01:46 no_c_quiz
20241026-22:01:46 weakest_accuracies [0.7776]
20241026-22:01:46 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-22:20:13 train_loss 9 model 0 0.040843696904182436
20241026-22:20:14 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:20:30 test_loss 9 model 0 0.028333854854106902
20241026-22:20:30 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:20:46 test_accuracy 9 model 0 nb_correct 2057 / 2500 (82.28%)
20241026-22:20:46 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-22:21:14 wrote ae_*.pth
20241026-22:21:14 epoch_duration 19min27s next_finish 22:40:42
20241026-22:21:14 wrote state.pth
20241026-22:21:14 --- epoch 10 ----------------------------------------
20241026-22:21:14 current_test_accuracies 0.8228
20241026-22:21:14 no_c_quiz
20241026-22:21:14 weakest_accuracies [0.8228]
20241026-22:21:14 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-22:39:41 train_loss 10 model 0 0.038202108073234556
20241026-22:39:41 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:39:57 test_loss 10 model 0 0.02720927083492279
20241026-22:39:57 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:40:14 test_accuracy 10 model 0 nb_correct 2180 / 2500 (87.20%)
20241026-22:40:14 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-22:40:43 wrote ae_*.pth
20241026-22:40:43 epoch_duration 19min28s next_finish 23:00:11
20241026-22:40:43 wrote state.pth
20241026-22:40:43 --- epoch 11 ----------------------------------------
20241026-22:40:43 current_test_accuracies 0.8720
20241026-22:40:43 no_c_quiz
20241026-22:40:43 weakest_accuracies [0.872]
20241026-22:40:43 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-22:59:14 train_loss 11 model 0 0.0375050094217062
20241026-22:59:14 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:59:30 test_loss 11 model 0 0.027711262971162796
20241026-22:59:30 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-22:59:46 test_accuracy 11 model 0 nb_correct 2259 / 2500 (90.36%)
20241026-22:59:46 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-23:00:17 wrote ae_*.pth
20241026-23:00:17 epoch_duration 19min33s next_finish 23:19:50
20241026-23:00:17 wrote state.pth
20241026-23:00:17 --- epoch 12 ----------------------------------------
20241026-23:00:17 current_test_accuracies 0.9036
20241026-23:00:17 no_c_quiz
20241026-23:00:17 weakest_accuracies [0.9036]
20241026-23:00:17 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-23:18:45 train_loss 12 model 0 0.034961151081323624
20241026-23:18:45 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-23:19:01 test_loss 12 model 0 0.024840730883181096
20241026-23:19:01 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-23:19:17 test_accuracy 12 model 0 nb_correct 2347 / 2500 (93.88%)
20241026-23:19:17 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-23:19:46 wrote ae_*.pth
20241026-23:19:46 epoch_duration 19min29s next_finish 23:39:15
20241026-23:19:46 wrote state.pth
20241026-23:19:46 --- epoch 13 ----------------------------------------
20241026-23:19:46 current_test_accuracies 0.9388
20241026-23:19:46 no_c_quiz
20241026-23:19:46 weakest_accuracies [0.9388]
20241026-23:19:46 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-23:38:14 train_loss 13 model 0 0.03403147121071815
20241026-23:38:14 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-23:38:30 test_loss 13 model 0 0.024604608163237573
20241026-23:38:30 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-23:38:46 test_accuracy 13 model 0 nb_correct 2358 / 2500 (94.32%)
20241026-23:38:46 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-23:39:15 wrote ae_*.pth
20241026-23:39:15 epoch_duration 19min28s next_finish 23:58:44
20241026-23:39:15 wrote state.pth
20241026-23:39:15 --- epoch 14 ----------------------------------------
20241026-23:39:15 current_test_accuracies 0.9432
20241026-23:39:15 no_c_quiz
20241026-23:39:15 weakest_accuracies [0.9432]
20241026-23:39:15 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241026-23:57:43 train_loss 14 model 0 0.03268781835436821
20241026-23:57:43 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-23:57:59 test_loss 14 model 0 0.031215623617172242
20241026-23:57:59 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241026-23:58:16 test_accuracy 14 model 0 nb_correct 2320 / 2500 (92.80%)
20241026-23:58:16 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241026-23:58:46 wrote ae_*.pth
20241026-23:58:46 epoch_duration 19min30s next_finish 00:18:16
20241026-23:58:46 wrote state.pth
20241026-23:58:46 --- epoch 15 ----------------------------------------
20241026-23:58:46 current_test_accuracies 0.9280
20241026-23:58:46 no_c_quiz
20241026-23:58:46 weakest_accuracies [0.928]
20241026-23:58:46 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241027-00:17:11 train_loss 15 model 0 0.031516154500842095
20241027-00:17:11 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241027-00:17:27 test_loss 15 model 0 0.025180682331323624
20241027-00:17:27 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241027-00:17:43 test_accuracy 15 model 0 nb_correct 2357 / 2500 (94.28%)
20241027-00:17:43 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241027-00:18:11 wrote ae_*.pth
20241027-00:18:11 epoch_duration 19min25s next_finish 00:37:37
20241027-00:18:11 wrote state.pth
20241027-00:18:11 --- epoch 16 ----------------------------------------
20241027-00:18:11 current_test_accuracies 0.9428
20241027-00:18:11 no_c_quiz
20241027-00:18:11 weakest_accuracies [0.9428]
20241027-00:18:11 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241027-00:36:36 train_loss 16 model 0 0.03089874274432659
20241027-00:36:36 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241027-00:36:52 test_loss 16 model 0 0.024968827426433565
20241027-00:36:52 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241027-00:37:08 test_accuracy 16 model 0 nb_correct 2293 / 2500 (91.72%)
20241027-00:37:08 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241027-00:37:37 wrote ae_*.pth
20241027-00:37:37 epoch_duration 19min25s next_finish 00:57:02
20241027-00:37:37 wrote state.pth
20241027-00:37:37 --- epoch 17 ----------------------------------------
20241027-00:37:37 current_test_accuracies 0.9172
20241027-00:37:37 no_c_quiz
20241027-00:37:37 weakest_accuracies [0.9172]
20241027-00:37:37 quiz_set nb_w_quizzes 50000 nb_c_quizzes 0
20241027-00:56:08 train_loss 17 model 0 0.030611307680606843
20241027-00:56:08 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241027-00:56:25 test_loss 17 model 0 0.02509162476658821
20241027-00:56:25 quiz_set nb_w_quizzes 2500 nb_c_quizzes 0
20241027-00:56:41 test_accuracy 17 model 0 nb_correct 2395 / 2500 (95.80%)
20241027-00:56:41 quiz_set nb_w_quizzes 150 nb_c_quizzes 0
20241027-00:57:10 wrote ae_*.pth
20241027-00:57:10 epoch_duration 19min33s next_finish 01:16:43
20241027-00:57:10 wrote state.pth
20241027-00:57:10 --- epoch 18 ----------------------------------------
20241027-00:57:10 current_test_accuracies 0.9580
20241027-00:57:14 wrote ae_*_naive.pth