-
Notifications
You must be signed in to change notification settings - Fork 29
/
training_run_notes_femur.txt
936 lines (707 loc) · 86.8 KB
/
training_run_notes_femur.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
real_good_robot runs
====================
Densenet push/grasp, trial reward
GPU 1, port ?, tab 3
commit: ?
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-01-28-15-01-52_Sim-Push-and-Grasp-Trial-Reward-Training
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --push_rewards --experience_replay --explore_rate_decay --save_visualizations --trial_reward --tcp_port 19987 --nn densenet
Densenet stack, trial reward
GPU 2, port 19998, tab 15
commit: b4756d9c51849dd0e2884acc7a109fc84c4335a2
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-01-28-15-05-06_Sim-Stack-Trial-Reward-Training
± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --trial_reward --tcp_port 19998 --nn densenet --place
Densenet stack, 2 step reward
GPU 1, port 19975, tab 16
commit: b4756d9c51849dd0e2884acc7a109fc84c4335a2
Densenet push/grasp 2 step reward, COMMON SENSE
GPU 0, port 19975, tab 0
commit: d50b5c64515dd68838685a419853bd10d23daccb
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19975 --common_sense --nn densenet --obj_mesh_dir objects/toys
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-01-29-14-59-43_Sim-Push-and-Grasp-Two-Step-Reward-Common-Sense-Training
--------------------- Below starts Jan 31
Densenet, rows, trial reward, common sense - JUNK, too many blocks
GPU 0, port 19975, tab 0
commit: 55cbdb8ea9eae684c54eb9bcfc361ae8fe9dbde0
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19975 --nn densenet --place --check_row --common_sense --trial_reward
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-01-31-18-02-20_Sim-Stack-Rows-Trial-Reward-Common-Sense-Training
Densenet, push/grasp, 2 step reward - SUPER BASIC BASELINE RUN
GPU 1, port 19987, tab 3
commit: 55cbdb8ea9eae684c54eb9bcfc361ae8fe9dbde0
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/toys --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19987 --nn densenet
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-01-31-18-06-45_Sim-Push-and-Grasp-Two-Step-Reward-Training
Densenet, rows, trial reward, common sense
GPU 0, port 19975, tab 0
commit: f7a98e8176c631d6ecc0a99d4727be7affac0d78
± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19975 --nn densenet --place --check_row --common_sense --trial_reward --num_obj 4
Feb 2 -------------------------------------------- MAJOR TRIAL REWARD BUGS FIXED FOR LAST TIMESTEP!!!
ROWS - No Common Sense
ahundt@femur|~/src/real_good_robot on fast_sim_thread
± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19975 --nn densenet --place --check_row --trial_reward --num_obj 4
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-02-20-30-18_Sim-Stack-Rows-Trial-Reward-Training
commit: 445c90a2bbf9b89c5c076231e8294c21126814e2
Tab 0, GPU 0, port 19975
Densenet, push/grasp, 2 step reward - SUPER BASIC BASELINE RUN
± export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/toys --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19987 --nn densenet
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-02-20-29-27_Sim-Push-and-Grasp-Two-Step-Reward-Training
commit: 445c90a2bbf9b89c5c076231e8294c21126814e2
Tab 3, GPU 1, port 19987
> TESTING PRESET CASES
> export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19999 --is_testing --random_seed 1238 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-02-20-29-27_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement.pth' --max_test_trials 10 --test_preset_cases
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-11-15-53-12_Sim-Push-and-Grasp-Two-Step-Reward-Testing
> Commit: 7b6c54ad615d592d86e71d90ea36c6478193a456
> Tab 1, GPU 1, port 19999
> TESTING CHALLENGING ARRANGEMENTS
> Max trial success rate: 0.9357798165137615, at action iteration: 1144. (total of 1146 actions, max excludes first 1144 actions)
> Max grasp success rate: 0.42344706911636043, at action iteration: 1145. (total of 1146 actions, max excludes first 1144 actions)
> Max grasp action efficiency: 0.4230769230769231, at action iteration: 1145. (total of 1147 actions, max excludes first 1144 actions)
> saving plot: 2020-02-11-15-53-12_Sim-Push-and-Grasp-Two-Step-Reward-Testing-Sim-Push-&-Grasp-VPG-Challenging-Arrangements_success_plot.png
> {'trial_success_rate_best_value': 0.9357798165137615, 'grasp_action_efficiency_best_value': 0.4230769230769231, 'grasp_success_rate_best_index': 1145, 'grasp_action_efficiency_best_index': 1145, 'trial_success_rate_best_index': 1144, 'grasp_success_rate_best_value': 0.42344706911636043}
> max trial successes: 103.0
> individual_arrangement_trial_success_rate: [0.9 0.9 0.9 0.9 1. 1. 0.6 1. 1. 1. 0.9]
> senarios_100_percent_complete: 5
>
> TEST RANDOM ARRANGEMENTS
> export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19999 --is_testing --random_seed 1238 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-02-20-29-27_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement.pth' --max_test_trials 100
> Pre-trained model snapshot loaded from: /home/ahundt/src/real_good_robot/logs/2020-02-02-20-29-27_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement.pth
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-14-20-52-58_Sim-Push-and-Grasp-Two-Step-Reward-Testing
Feb 3 --------------------------------------------- MAJOR TRIAL REWARD CHANGE DOUBLE CREDIT LAST TIMESTEP!!!
Densenet, push/grasp, trial_reward - Trial Reward BASELINE RUN
± export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/toys --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19987 --nn densenet --trial_reward
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-03-15-43-21_Sim-Push-and-Grasp-Trial-Reward-Training
commit: 4911dbee967553d6447d83e8053c6acc2bfe7a07
Tab 3, GPU 1, port 19987
ROWS- Common Sense
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19975 --nn densenet --place --check_row --trial_reward --num_obj 4 --common_sense
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-03-16-02-46_Sim-Rows-Trial-Reward-Common-Sense-Training
commit: 4911dbee967553d6447d83e8053c6acc2bfe7a07
Tab 0, GPU 0, port 19975
PixelNet Debugging - DenseNet - push/grasp - 2 step reward - looks OK!
± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/toys --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19998 --nn densenet
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-03-17-35-43_Sim-Push-and-Grasp-Two-Step-Reward-Training
Tab 6, GPU 2, port 19998
PixelNet Debugging - efficientnet - push/grasp - 2 step reward
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/toys --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19998 --nn efficientnet
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-04-17-26-32_Sim-Push-and-Grasp-Two-Step-Reward-Training
Tab 6, GPU 2, port 19998
2019-02-10
Rows + common sense + densenet + trial reward, note: future reward discount rate is default of 0.5
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19965 --nn densenet --place --check_row --trial_reward --num_obj 4 --common_sense
commit: 7b6c54ad615d592d86e71d90ea36c6478193a456
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-10-18-38-48_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Tab 0, GPU 0, port 19965
> TESTING
> export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19965 --nn densenet --place --check_row --trial_reward --num_obj 4 --common_sense --is_testing --random_seed 1238 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-10-18-38-48_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training/models/snapshot.reinforcement-best-stack-rate.pth' --max_test_trials 100
> Commit: b2222ca2db65c0d5571b0afd428b1ff53013c60d
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-14-15-24-00_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing
> video: recording_2020_02_14-15_23-46.avi
> Max trial success rate: 0.9292929292929293, at action iteration: 1086. (total of 1088 actions, max excludes first 1086 actions)
> Max grasp success rate: 0.8660550458715597, at action iteration: 1087. (total of 1088 actions, max excludes first 1086 actions)
> Max action efficiency: 0.856353591160221, at action iteration: 1086. (total of 1089 actions, max excludes first 1086 actions)
> saving plot: 2020-02-14-15-24-00_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing-Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing_success_plot.png
> {'action_efficiency_best_index': 1086, 'grasp_success_rate_best_value': 0.8660550458715597, 'place_success_rate_best_index': None, 'trial_success_rate_best_index': 1086, 'trial_success_rate_best_value': 0.92929292929292
93, 'grasp_success_rate_best_index': 1087, 'place_success_rate_best_value': -inf, 'action_efficiency_best_value': 0.856353591160221}
> Pre-trained model snapshot loaded from: /home/ahundt/src/real_good_robot/logs/2020-02-10-18-38-48_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training/models/snapshot.reinforcement-best-stack-rate.pth
> RUN WITH MODELS NOT RELOADING, DO NOT USE: d02a77736b978cda32f86b9bf018a639894c1a09
> RUN WITH MODELS NOT RELOADING, DO NOT USE: Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-12-21-10-24_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing
> RUN WITH MODELS NOT RELOADING, DO NOT USE: video: recording_2020_02_12-21_12-16.avi
> Tab 0, GPU 0, port 19965
2019-02-11
push + grasp + common sense + efficient net + trial reward
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir 'objects/blocks' --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --place --tcp_port 19961 --common_sense --trial_reward --future_reward_discount 0.65 --nn efficientnet --check_z_height
commit: 7b6c54ad615d592d86e71d90ea36c6478193a456
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-11-15-36-41_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Tab 2, GPU 2, port 19961
2019-02-13
Rows + common sense + densenet + trial reward, CRITICAL BUGFIX ON EXPERIENCE REPLAY FOR PLACE ACTION
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19965 --nn densenet --plac│·····················································
e --check_row --trial_reward --num_obj 4 --common_sense --future_reward_discount 0.65
commit: 2b55d4b48c2c6fa1959e52947691b26355aa4180
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-13-18-26-52_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Tab 0, GPU 0, port 19965
2019-02-14
Rows + common sense + densenet + trial reward, CRITICAL BUGFIX ON EXPERIENCE REPLAY FOR PLACE ACTION, and plotting
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --push_rewards --experience_replay --explore_rate_decay --save_visualizations --tcp_port 19965 --nn densenet --place --check_row --trial_reward --num_obj 4 --common_sense --future_reward_discount 0.65
commit: 4666ac42e9f474ad51a352212134dffa87918ddf
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-14-20-48-19_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Tab 0, GPU 0, port 19965
Feb 16 --------------------------------------------- INGEGRATED TRAIN VAL TEST RUNS THESE RESULTS ARE IN THE PAPER!!!
push + grasp, common sense, densenet, trial reward, 5000 actions
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19961 --common_sense --trial_reward --future_reward_discount 0.65 --nn densenet --max_train_actions 5000
commit: a34337edf89e77843bb11e1618666e5586e072f3
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-59_Sim-Push-and-Grasp-SPOT-Trial-Reward-Common-Sense-Training
Random Arrangement Testing Results: {"grasp_action_efficiency_best_value": 0.74, "trial_success_rate_best_value": 1.0, "trial_success_rate_best_index": 633, "grasp_success_rate_best_index": 4859, "grasp_action_efficiency_best_index": 4991, "grasp_success_rate_best_value": 0.8360655737704918}
> Challenging Arrangement Testing
> Commit: 9e055205b738be1ed60a189f18e2362c6603f331
> export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19961 --common_sense --trial_reward --future_reward_discount 0.65 --nn densenet --random_seed 1238 --save_visualizations --is_testing --test_preset_cases --max_test_trials 10 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-59_Sim-Push-and-Grasp-SPOT-Trial-Reward-Common-Sense-Training/models/snapshot.reinforcement_grasp_action_efficiency_best_value.pth'
> Pre-trained model snapshot loaded from: /home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-59_Sim-Push-and-Grasp-SPOT-Trial-Reward-Common-Sense-Training/models/snapshot.reinforcement_grasp_action_efficiency_best_value.pth
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-19-22-05-10_Sim-Push-and-Grasp-SPOT-Trial-Reward-Common-Sense-Challenging-Arrangements
> Video: recording_2020_02_19-22_04-55.avi
Tab 0, GPU 0, port 19961
push + grasp, densenet, trial reward, 5000 actions
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19975 --trial_reward --future_reward_discount 0.65 --nn densenet --max_train_actions 5000
commit: a34337edf89e77843bb11e1618666e5586e072f3
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-16-21-37-47_Sim-Push-and-Grasp-SPOT-Trial-Reward-Training
Random Arrangement Testing Results: {"grasp_action_efficiency_best_index": 2924, "grasp_success_rate_best_value": 0.8341346153846154, "trial_success_rate_best_index": 608, "trial_success_rate_best_value": 1.0, "grasp_success_rate_best_index": 3872, "grasp_action_efficiency_best_value": 0.736}
> Challenging Arrangement Testing
> Commit: 9e055205b738be1ed60a189f18e2362c6603f331
> export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19975 --trial_reward --future_reward_discount 0.65 --nn densenet --random_seed 1238 --save_visualizations --is_testing --test_preset_cases --max_test_trials 10 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-16-21-37-47_Sim-Push-and-Grasp-SPOT-Trial-Reward-Training/models/snapshot.reinforcement_grasp_action_efficiency_best_value.pth'
> Pre-trained model snapshot loaded from: /home/ahundt/src/real_good_robot/logs/2020-02-16-21-37-47_Sim-Push-and-Grasp-SPOT-Trial-Reward-Training/models/snapshot.reinforcement_grasp_action_efficiency_best_value.pth
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-19-22-07-44_Sim-Push-and-Grasp-SPOT-Trial-Reward-Challenging-Arrangements
> Video: recording_2020_02_19-22_07-30.avi
Tab 1, GPU 1, port 19975
push + grasp, densenet, 5000 actions -- SUPER BASIC PUSH GRASP RUN
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19999 --future_reward_discount 0.65 --nn densenet --max_train_actions 5000
commit: a34337edf89e77843bb11e1618666e5586e072f3
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-55_Sim-Push-and-Grasp-Two-Step-Reward-Training
Random Arrangement Testing Results: {"trial_success_rate_best_value": 0.9393939393939394, "trial_success_rate_best_index": 1403, "grasp_action_efficiency_best_index": 1403, "grasp_success_rate_best_index": 1403, "grasp_success_rate_best_value": 0.7877280265339967, "grasp_action_efficiency_best_value": 0.67712045616536}
> Challenging Arrrangement Testing (Something is odd about this one?)
> Commit: 9e055205b738be1ed60a189f18e2362c6603f331
> export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19999 --future_reward_discount 0.65 --nn densenet --random_seed 1238 --save_visualizations --is_testing --test_preset_cases --max_test_trials 10 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-55_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement_grasp_action_efficiency_best_value.pth'
> Pre-trained model snapshot loaded from: /home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-55_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement_grasp_action_efficiency_best_value.pth
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-19-22-11-11_Sim-Push-and-Grasp-Two-Step-Reward-Challenging-Arrangements
> Video: recording_2020_02_19-22_10-47.avi
> Results: {"grasp_success_rate_best_index": 1412, "grasp_action_efficiency_best_index": 1412, "trial_success_rate_best_index": null, "grasp_action_efficiency_best_value": 0.1990084985835694, "grasp_success_rate_best_value": 0.5597609561752988, "senarios_100_percent_complete": 2, "trial_success_rate_best_value": -Infinity}
>
> Random Arrangements Testing V2
> export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19975 --future_reward_discount 0.65 --nn densenet --random_seed 1238 --save_visualizations --is_testing --max_test_trials 100 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-55_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement_grasp_success_rate_best_value.pth'
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-20-15-12-36_Sim-Push-and-Grasp-Two-Step-Reward-Testing
> Video: recording_2020_02_20-15_12-24.avi
> Results: {'trial_success_rate_best_index': 1293, 'grasp_success_rate_best_value': 0.766637856525497, 'trial_success_rate_best_value': 0.8585858585858586, 'grasp_action_efficiency_best_value': 0.6867749419953596, 'grasp_action_efficiency_best_index': 1295, 'grasp_success_rate_best_index': 1293}
> Tab 1, GPU 1, port 19975
>
> Challenging Arrangement Testing V2
> export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir 'objects/toys' --num_obj 10 --push_rewards --experience_replay --explore_rate_decay --tcp_port 19999 --future_reward_discount 0.65 --nn densenet --random_seed 1238 --save_visualizations --is_testing --test_preset_cases --max_test_trials 10 --snapshot_file '/home/ahundt/src/real_good_robot/logs/2020-02-16-21-33-55_Sim-Push-and-Grasp-Two-Step-Reward-Training/models/snapshot.reinforcement_grasp_success_rate_best_value.pth'
> Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-20-15-13-34_Sim-Push-and-Grasp-Two-Step-Reward-Challenging-Arrangements
> Video: recording_2020_02_20-15_13-25.avi
> Results: {"grasp_action_efficiency_best_index": 1124, "grasp_action_efficiency_best_value": 0.3736654804270463, "grasp_success_rate_best_index": 1124, "grasp_success_rate_best_value": 0.498812351543943, "senarios_100_percent_complete": 5, "trial_success_rate_best_index": 1124, "trial_success_rate_best_value": 0.908256880733945}
> Tab 2, GPU 2, port 19999
Tab 2, GPU 2, port 19999
Feb 20 ----------------------- 2020-02-20
Stacking, Mask, no backprop, densenet, trial reward -- FOR PAPER
================================================================
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --save_visualizations --common_sense --check_z_height --tcp_port 19961 --place --future_reward_discount 0.65 --max_train_actions 10000 --no_common_sense_backprop
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-02-20-16-20-23_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Commit: e6363c7248adce5ddab1c2409d1b90bb86e08dab
Tab 0, GPU 0, port 19961
=============================================================
2020-04 and 2020-05
=============================================================
Tab 7: ~/src/V-REP_PRO_EDU_V3_6_2_Ubuntu16_04/vrep.sh -gREMOTEAPISERVERSERVICE_19965_FALSE_TRUE -s ~/src/real_good_robot/simulation/simulation.ttt
Tab 8: ~/src/V-REP_PRO_EDU_V3_6_2_Ubuntu16_04/vrep.sh -gREMOTEAPISERVERSERVICE_19999_FALSE_TRUE -s simulation/simulation.ttt
IGNORE (crashed out early) - SIM STACK - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN
-----------------------------------------------------------
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --save_visualizations --common_sense --check_z_height --tcp_port 19990 --place --future_reward_discount 0.65 --max_train_actions 10000 --tcp_port 19965
RESUME: export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --save_visualizations --common_sense --check_z_height --tcp_port 19990 --place --future_reward_discount 0.65 --max_train_actions 10000 --tcp_port 19965 --resume /home/ahundt/src/real_good_robot/logs/2020-04-27-10-23-17_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-04-27-10-23-17_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
RUN CRASHED, NEEDED TO RESTART: Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-04-26-16-03-44_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Commit: b11dc983b93e1b3cf4beeef7a786b51e9df0f751
GPU 0, Tab 0, port 19965, left v-rep window, v-rep tab 7
IGNORE (crashed out early) - SIM ROW - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN
-----------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --trial_reward --save_visualizations --common_sense --check_row --tcp_port 19999 --place --future_reward_discount 0.65 --max_train_actions 10000
RESUME: export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --trial_reward --save_visualizations --common_sense --check_row --tcp_port 19999 --place --future_reward_discount 0.65 --max_train_actions 10000 --resume /home/ahundt/src/real_good_robot/logs/2020-04-26-16-09-48_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-04-26-16-09-48_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Commit: b11dc983b93e1b3cf4beeef7a786b51e9df0f751
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
SIM STACK - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - TRULY RANDOM ACTION EXPLORATION
---------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 19965 --future_reward_discount 0.65 --max_train_actions 10000
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-03-15-29-17_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
RESUME: export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 19965 --future_reward_discount 0.65 --max_train_actions 10000 --resume /home/ahundt/src/real_good_robot/logs/2020-05-03-15-29-17_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Commit: ce986ed39086d1e705f9d30325586c05fb1db469
Random Testing results:
{'action_efficiency_best_value': 0.49328859060402686, 'grasp_success_rate_best_value': 0.7327459618208517, 'trial_success_rate_best_index': 1192, 'trial_success_rate_best_value':
0.97, 'grasp_success_rate_best_index': 1193, 'place_success_rate_best_index': 1194, 'action_efficiency_best_index': 1194, 'place_success_rate_best_value': 0.7914230019493177}
Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-03-15-29-17_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Training results:
{'action_efficiency_best_value': 0.456, 'grasp_success_rate_best_value': 0.7857142857142857, 'trial_success_rate_best_index': 8835, 'trial_success_rate_best_value': 0.692307692307
6923, 'grasp_success_rate_best_index': 8952, 'place_success_rate_best_index': 9987, 'action_efficiency_best_index': 9389, 'place_success_rate_best_value': 0.7688679245283019}
GPU 0, Tab 0, port 19965, left v-rep window, v-rep tab 7
SIM ROW - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - TRULY RANDOM ACTION EXPLORATION
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_row --tcp_port 19999 --place --future_reward_discount 0.65 --max_train_actions 10000
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-03-20-04-47_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
FILE CORRUPTED: Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-03-15-29-18_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Commit: ce986ed39086d1e705f9d30325586c05fb1db469
Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-03-20-04-47_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training/2020-05-06-08-57-05_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing
Random Testing results:
{'grasp_success_rate_best_value': 0.0, 'action_efficiency_best_index': 1815, 'place_success_rate_best_index': 1815, 'place_success_rate_best_value': 0.0, 'trial_success_rate_best_index': 1815, 'trial_success_rate_best_value': 0.13, 'grasp_success_rate_best_index': 1815, 'action_efficiency_best_value': 0.0}
Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-03-20-04-47_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Training results:
{'grasp_success_rate_best_value': 0.75, 'action_efficiency_best_index': 814, 'place_success_rate_best_index': 6683, 'place_success_rate_best_value': 1.0, 'trial_success_rate_best_index': 9523, 'trial_success_rate_best_value': 0.175, 'grasp_success_rate_best_index': 8617, 'action_efficiency_best_value': 0.132}
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
SIM ROW - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - TRULY RANDOM ACTION EXPLORATION - Efficientnet
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 10000 --nn efficientnet
Commit: ce986ed39086d1e705f9d30325586c05fb1db469 + manually switch reinforcement_net -> PixelNet
saving plot: 2020-05-07-04-41-58_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing-Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing_success_plot.png
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-07-04-41-58_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing/data/best_stats.json
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-07-04-41-58_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing/best_stats.json
Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-04-12-08-15_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training/2020-05-07-04-41-58_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing
Random Testing results:
{'action_efficiency_best_value': 0.39863013698630134, 'place_success_rate_best_index': 1460, 'place_success_rate_best_value': 0.7187039764359352, 'action_efficiency_best_index': 1462, 'grasp_success_rate_best_value': 0.7979539641943734, 'trial_success_rate_best_value': 0.95, 'trial_success_rate_best_index': 1460, 'grasp_success_rate_best_index': 1460}
Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-04-12-08-15_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Training results:
{'action_efficiency_best_value': 0.42, 'place_success_rate_best_index': 9057, 'place_success_rate_best_value': 0.7424892703862661, 'action_efficiency_best_index': 8867, 'grasp_success_rate_best_value': 0.8812260536398467, 'trial_success_rate_best_value': 0.5645161290322581, 'trial_success_rate_best_index': 8939, 'grasp_success_rate_best_index': 8895}
ahundt@femur|~/src/real_good_robot on fast_sim_thread!?
± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 10000 --nn efficientnet --disable_two_step_backprop -> [1]
ahundt@femur|~/src/real_good_robot on fast_sim_thread!?
± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 10000 --nn efficientnet --disable_two_step_backprop --random_actions --resume '/home/ahundt/src/real_good_robot/logs/2020-05-04-12-08-15_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training'
GPU 2, Tab 1, port 20000, right v-rep window, v-rep tab 9
SIM STACK - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.125, 1, 1 - femur 2020-05-06
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 19965 --future_reward_discount 0.65 --max_train_actions 10000 --random_actions
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-06-20-53-21_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Commit: 1c0359b86df7553dd01343d2a31a4f88ddbb41fd
GPU 0, Tab 0, port 19965, left v-rep window, v-rep tab 7
TODO: rerun testing 2020-05-09-10-16-34_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing-Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing_success_plot,
the simulator got into a bad state during the test evaluation
SIM ROW - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - Mixed RANDOM ACTION, 2D ACTION - REWARD SCHEDULE 0.125, 1, 1 - femur 2020-05-06
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_row --tcp_port 19999 --place --future_reward_discount 0.65 --max_train_actions 10000 --random_actions
RESUME: export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_row --tcp_port 19999 --place --future_reward_discount 0.65 --max_train_actions 10000 --random_actions --resume /home/ahundt/src/real_good_robot/logs/2020-05-06-20-58-40_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-06-20-58-40_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Commit: 1c0359b86df7553dd01343d2a31a4f88ddbb41fd
Commit (resume fix row check): c6c4b401fe719aae89966adaf9ed5ca24cf95fde
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
saving plot: 2020-05-09-22-03-51_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing-Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing_success_plot.png
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-09-22-03-51_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing/data/best_stats.json
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-09-22-03-51_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing/best_stats.json
Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-06-20-58-40_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training/2020-05-09-22-03-51_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Testing
Random Testing results:
{'trial_success_rate_best_index': 1216, 'grasp_success_rate_best_value': 0.6693989071038251, 'place_success_rate_best_value': 0.7818930041152263, 'action_efficiency_best_index': 1218, 'place_success_rate_best_index': 1216, 'trial_success_rate_best_value': 0.94, 'grasp_success_rate_best_index': 1216, 'action_efficiency_best_value': 0.5180921052631579}
Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-06-20-58-40_Sim-Rows-SPOT-Trial-Reward-Common-Sense-Training
Training results:
{'trial_success_rate_best_index': 6948, 'grasp_success_rate_best_value': 0.5954692556634305, 'place_success_rate_best_value': 0.8058823529411765, 'action_efficiency_best_index': 9733, 'place_success_rate_best_index': 9702, 'trial_success_rate_best_value': 0.4230769230769231, 'grasp_success_rate_best_index': 6326, 'action_efficiency_best_value': 0.576}
DO NOT USE FOR FINAL RESULTS - SIM STACK - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - EFFICIENTNET, no dilation - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-07
---------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --nn efficientnet --random_actions
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-07-14-52-29_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Commit: 7be259d4b37e41a7e0bb7900aaf6e0220d336f3b
GPU 2, Tab 1, port 20000, right v-rep window, v-rep tab 9
Why do not use for final results?
Simulator got into a bad state (probably physics stuck object) around 5000 actions in and around 7000 actions in.
SIM STACK - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - EFFICIENTNET, no dilation - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-09
---------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --nn efficientnet --random_actions
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-09-15-02-34_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Commit: 7be259d4b37e41a7e0bb7900aaf6e0220d336f3b
GPU 2, Tab 1, port 20000, right v-rep window, v-rep tab 9
STACK: trial: 101 actions/partial: 3.6656626506024095 actions/full stack: 12.292929292929292 (lower is better) Grasp Count: 713, grasp success rate: 0.7068723702664796 place_on_stack_rate: 0.6600397614314115 place_attempts: 503 partial_stack_successes: 332 stack_successes: 99 trial_success_rate: 0.9801980198019802 stack goal: None current_height: 1.0225278559120623
trial_complete_indices: [ 9. 15. 25. 34. 41. 45. 51. 66. 79. 85. 110. 114.
120. 129. 133. 145. 153. 169. 176. 186. 192. 202. 220. 227.
234. 240. 264. 270. 282. 288. 297. 305. 314. 327. 335. 341.
347. 354. 366. 379. 387. 393. 400. 423. 429. 435. 453. 463.
469. 492. 538. 558. 574. 580. 588. 597. 603. 609. 616. 629.
637. 645. 651. 657. 665. 674. 768. 779. 793. 798. 805. 816.
941. 948. 970. 979. 988. 996. 1008. 1027. 1035. 1047. 1053. 1061.
1084. 1090. 1098. 1104. 1114. 1124. 1131. 1137. 1143. 1156. 1163. 1170.
1183. 1190. 1200. 1206. 1216.]
Max trial success rate: 0.98, at action iteration: 1213. (total of 1215 actions, max excludes first 1213 actions)
Max grasp success rate: 0.7088607594936709, at action iteration: 1214. (total of 1215 actions, max excludes first 1213 actions)
Max place success rate: 0.8115079365079365, at action iteration: 1215. (total of 1216 actions, max excludes first 1213 actions)
Max action efficiency: 0.4896949711459192, at action iteration: 1215. (total of 1216 actions, max excludes first 1213 actions)
saving plot: 2020-05-14-02-09-24_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing-Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing_success_plot.png
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-14-02-09-24_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing/data/best_stats.json
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-14-02-09-24_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing/best_stats.json
Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-09-15-02-34_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training/2020-05-14-02-09-24_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Testing
Random Testing results:
{'grasp_success_rate_best_index': 1214, 'place_success_rate_best_index': 1215, 'trial_success_rate_best_value': 0.98, 'trial_success_rate_best_index': 1213, 'action_efficiency_best_index': 1215, 'action_efficiency_best_value': 0.4896949711459192, 'grasp_success_rate_best_value': 0.7088607594936709, 'place_success_rate_best_value': 0.8115079365079365}
Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-09-15-02-34_Sim-Stack-SPOT-Trial-Reward-Common-Sense-Training
Training results:
{'grasp_success_rate_best_index': 19667, 'place_success_rate_best_index': 18326, 'trial_success_rate_best_value': 0.7704918032786885, 'trial_success_rate_best_index': 19962, 'action_efficiency_best_index': 9143, 'action_efficiency_best_value': 0.576, 'grasp_success_rate_best_value': 0.8345588235294118, 'place_success_rate_best_value': 0.8509615384615384}
SIM STACK - COMMON SENSE - DISCOUNTED REWARD - WITH SPOT-Q - FULL FEATURED RUN - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-06
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --discounted_reward --common_sense --check_z_height --place --tcp_port 19965 --future_reward_discount 0.9 --max_train_actions 20000 --random_actions --disable_two_step_backprop
RESUME: ± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --discounted_reward --common_sense --check_z_height --place --tcp_port 19965 --future_reward_discount 0.9 --max_train_actions 20000 --random_actions --disable_two_step_backprop --resume /home/ahundt/src/real_good_robot/logs/2020-05-12-14-41-34_Sim-Stack-Two-Step-Reward-Common-Sense-Training
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-12-14-41-34_Sim-Stack-Two-Step-Reward-Common-Sense-Training
Commit: 4b9401fb69c809e8230b9d2f4b3627e5329a507d
Resume Commit: 764e4f6e9bf3b33943640dd8e1fb5984faf783d0
GPU 0, Tab 0, port 19965, left v-rep window, v-rep tab 7
> Testing results (abbreviated, 49 trials): TODO(ahundt) consider resuming again, but this should be good enough for the table. Stopped to do other runs where the results are more critical.
> Testing iteration: 1166
> prev_height: 0.0 max_z: 0.05115742769372662 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> Current count of pixels with stuff: 5083.0 threshold below which the scene is considered empty: 10
> Change detected: True (value: 6122)
> Trainer.get_label_value(): Current reward: 0.000000 Current reward multiplier: 1.024000 Predicted Future reward: 0.627358 Expected reward: 0.000000 + 0.900000 x 0.627358 = 0.564622
> trial_complete_indices: [ 14. 26. 41. 77. 89. 102. 115. 161. 188. 218. 254. 279.
> 296. 311. 326. 338. 360. 397. 422. 437. 479. 501. 525. 537.
> 597. 613. 627. 644. 684. 732. 760. 811. 823. 835. 854. 871.
> 913. 946. 960. 978. 990. 1012. 1027. 1055. 1067. 1098. 1110. 1139.
> 1166.]
> Max trial success rate: 0.0, at action iteration: 1163. (total of 1165 actions, max excludes first 1163 actions)
> Max grasp success rate: 0.17142857142857143, at action iteration: 1163. (total of 1165 actions, max excludes first 1163 actions)
> Max place success rate: 0.3710247349823322, at action iteration: 1163. (total of 1166 actions, max excludes first 1163 actions)
> Max action efficiency: 0.0, at action iteration: 1163. (total of 1166 actions, max excludes first 1163 actions)
> saving plot: 2020-05-18-14-30-08_Sim-Stack-Discounted-Reward-Masked-Testing-Sim-Stack-Discounted-Reward-Masked-Testing_success_plot.png
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-18-14-30-08_Sim-Stack-Discounted-Reward-Masked-Testing/data/best_stats.json
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-18-14-30-08_Sim-Stack-Discounted-Reward-Masked-Testing/best_stats.json
> Trial logging complete: 49 --------------------------------------------------------------
> Primitive confidence scores: 0.418146 (push), 0.408318 (grasp), 0.627358 (place)
> Action: push at (6, 120, 183)
SIM STACK - COMMON SENSE - DISCOUNTED REWARD - NO SPOT-Q - FULL FEATURED RUN - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-06
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --discounted_reward --check_z_height --place --tcp_port 19999 --future_reward_discount 0.9 --max_train_actions 20000 --random_actions --disable_two_step_backprop
RESUME: ± export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --discounted_reward --check_z_height --place --tcp_port 19999 --future_reward_discount 0.9 --max_train_actions 20000 --random_actions --disable_two_step_backprop --resume /home/ahundt/src/real_good_robot/logs/2020-05-12-16-47-11_Sim-Stack-Discounted-Reward-Training
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-12-16-47-11_Sim-Stack-Discounted-Reward-Training
Commit: 760c5db15551001227424b64b35d4e852a5ec74e
Resume Commit: 764e4f6e9bf3b33943640dd8e1fb5984faf783d0
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
> Testing results:
> Trial logging complete: 101 --------------------------------------------------------------
> Primitive confidence scores: 0.111208 (push), 0.049239 (grasp), 0.515781 (place)
> Action: push at (15, 7, 23)
> Predicting push action failure, heuristics determined push at height 0.00099447991561602 would not contact anything at the max height of 0.00109729329212217
> Executing: Push at (-0.678000, -0.210000, 0.000994) angle: 5.890486
> gripper position: 0.03216123580932617
> gripper position: 0.026226460933685303
> gripper position: 0.001162111759185791
> gripper position: -0.023666560649871826
> gripper position: -0.041889071464538574
> prev_height: 0.0 max_z: 0.0511317473084469 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> prev_height: 1.0 max_z: 1.022634946168938 goal_success: False needed to reset: False max_workspace_height: 0.6 <<<<<<<<<<<
> check_stack() stack_height: 1.022634946168938 stack matches current goal: False partial_stack_success: False Does the code think a reset is needed: False
> Push motion successful (no crash, need not move blocks): True
> STACK: trial: 101 actions/partial: inf actions/full stack: inf (lower is better) Grasp Count: 405, grasp success rate: 0.0049382716049382715 place_on_stack_rate: 0 place_attempts: 2 partial_stack_successes: 0 stack_successes: 0 trial_success_rate: inf stack goal: None current_height: 1.022634946168938
> trial_complete_indices: [ 12. 24. 37. 65. 77. 89. 101. 114. 126. 140. 152. 165.
> 184. 196. 212. 227. 242. 254. 266. 279. 291. 303. 322. 335.
> 352. 373. 386. 398. 410. 422. 436. 452. 468. 480. 493. 505.
> 518. 540. 553. 565. 584. 596. 608. 621. 636. 648. 660. 672.
> 684. 696. 708. 720. 732. 744. 756. 768. 780. 804. 821. 833.
> 845. 857. 869. 881. 893. 909. 921. 933. 945. 957. 970. 992.
> 1004. 1016. 1029. 1041. 1053. 1065. 1077. 1089. 1101. 1113. 1126. 1138.
> 1150. 1163. 1175. 1187. 1199. 1211. 1223. 1236. 1248. 1260. 1272. 1284.
> 1296. 1308. 1320. 1332. 1345.]
> Max trial success rate: 0.0, at action iteration: 1342. (total of 1344 actions, max excludes first 1342 actions)
> Max grasp success rate: 0.0049504950495049506, at action iteration: 1343. (total of 1344 actions, max excludes first 1342 actions)
> Max place success rate: 0.35074626865671643, at action iteration: 1342. (total of 1345 actions, max excludes first 1342 actions)
> Max action efficiency: 0.0, at action iteration: 1342. (total of 1345 actions, max excludes first 1342 actions)
> saving plot: 2020-05-17-18-59-57_Sim-Stack-Discounted-Reward-Testing-Sim-Stack-Discounted-Reward-Testing_success_plot.png
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-17-18-59-57_Sim-Stack-Discounted-Reward-Testing/data/best_stats.json
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-17-18-59-57_Sim-Stack-Discounted-Reward-Testing/best_stats.json
> Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-12-16-47-11_Sim-Stack-Discounted-Reward-Training/2020-05-17-18-59-57_Sim-Stack-Discounted-Reward-TestingRandom Testing results:
> {'action_efficiency_best_value': 0.0, 'action_efficiency_best_index': 1342, 'grasp_success_rate_best_index': 1343, 'place_success_rate_best_value': 0.35074626865671643, 'place_success_rate_best_index': 1342, 'trial_success_rate_best_value': 0.0, 'grasp_success_rate_best_value': 0.0049504950495049506, 'trial_success_rate_best_index': 1342}
> Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-12-16-47-11_Sim-Stack-Discounted-Reward-Training
> Training results:
> {'action_efficiency_best_value': 0.0, 'action_efficiency_best_index': 500, 'grasp_success_rate_best_index': 12664, 'place_success_rate_best_value': 0.46258503401360546, 'place_success_rate_best_index': 13309, 'trial_success_rate_best_value': 0.0, 'grasp_success_rate_best_value': 0.05090909090909091, 'trial_success_rate_best_index': 500}
SIM STACK - COMMON SENSE - TRIAL REWARD - FULL FEATURED RUN - EFFICIENTNET, 1 dilation - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-09
---------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --nn efficientnet --num_dilation 1 --random_actions
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-15-09-59-06_Sim-Stack-SPOT-Trial-Reward-Masked-Training
Resume: ± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --trial_reward --common_sense --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --nn efficientnet --num_dilation 1 --random_actions --resume /home/ahundt/src/real_good_robot/logs/2020-05-15-09-59-06_Sim-Stack-SPOT-Trial-Reward-Masked-Training
Commit: adb9792b15704fb96adf97622235d6547cbf8386
GPU 2, Tab 1, port 20000, right v-rep window, v-rep tab 9
Max trial success rate: 0.98, at action iteration: 807. (total of 809 actions, max excludes first 807 actions)
Max grasp success rate: 0.9033018867924528, at action iteration: 807. (total of 809 actions, max excludes first 807 actions)
Max place success rate: 0.84375, at action iteration: 807. (total of 810 actions, max excludes first 807 actions)
Max action efficiency: 0.7360594795539034, at action iteration: 809. (total of 810 actions, max excludes first 807 actions)
saving plot: 2020-05-21-22-35-13_Sim-Stack-SPOT-Trial-Reward-Masked-Testing-Sim-Stack-SPOT-Trial-Reward-Masked-Testing_success_plot.png
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-21-22-35-13_Sim-Stack-SPOT-Trial-Reward-Masked-Testing/data/best_stats.json
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-21-22-35-13_Sim-Stack-SPOT-Trial-Reward-Masked-Testing/best_stats.json
Trial logging complete: 101 --------------------------------------------------------------
Running two step backprop()
Primitive confidence scores: 1.149409 (push), 4.954260 (grasp), 10.060949 (place)
Action: grasp at (0, 131, 131)
Training loss: 0.000013
Executing: grasp at (-0.462000, 0.038000, 0.001001) orientation: 0.000000
gripper position: 0.023079276084899902
Grasp successful: False
prev_height: 0.0 max_z: 0.06238639676919751 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
prev_height: 1.0 max_z: 1.24772793538395 goal_success: False needed to reset: False max_workspace_height: 0.6 <<<<<<<<<<<
check_stack() stack_height: 1.24772793538395 stack matches current goal: False partial_stack_success: False Does the code think a reset is needed: False
STACK: trial: 101 actions/partial: 2.6765676567656764 actions/full stack: 8.191919191919192 (lower is better) Grasp Count: 426, grasp success rate: 0.9014084507042254 place_on_stack_rate: 0.7911227154046997 place_attempts: 383 partial_stack_successes: 303 stack_successes: 99 trial_success_rate: 0.9801980198019802 stack goal: None current_height: 1.24772793538395
trial_complete_indices: [ 6. 12. 20. 28. 42. 47. 55. 69. 78. 85. 97. 101. 108. 115.
123. 130. 138. 142. 148. 154. 161. 173. 179. 188. 196. 202. 206. 216.
222. 228. 234. 240. 246. 252. 261. 265. 295. 305. 313. 327. 337. 341.
347. 355. 367. 375. 379. 385. 391. 401. 407. 417. 423. 429. 435. 441.
447. 461. 469. 473. 481. 487. 491. 499. 507. 515. 521. 529. 537. 541.
550. 556. 560. 568. 574. 589. 597. 627. 631. 637. 643. 649. 657. 663.
669. 677. 683. 691. 702. 709. 731. 737. 741. 757. 763. 770. 775. 783.
789. 795. 810.]
Max trial success rate: 0.98, at action iteration: 807. (total of 809 actions, max excludes first 807 actions)
Max grasp success rate: 0.9033018867924528, at action iteration: 807. (total of 809 actions, max excludes first 807 actions)
Max place success rate: 0.84375, at action iteration: 807. (total of 810 actions, max excludes first 807 actions)
Max action efficiency: 0.7360594795539034, at action iteration: 809. (total of 810 actions, max excludes first 807 actions)
saving plot: 2020-05-21-22-35-13_Sim-Stack-SPOT-Trial-Reward-Masked-Testing-Sim-Stack-SPOT-Trial-Reward-Masked-Testing_success_plot.png
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-21-22-35-13_Sim-Stack-SPOT-Trial-Reward-Masked-Testing/data/best_stats.json
saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-21-22-35-13_Sim-Stack-SPOT-Trial-Reward-Masked-Testing/best_stats.json
Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-15-09-59-06_Sim-Stack-SPOT-Trial-Reward-Masked-Training/2020-05-21-17-27-35_Sim-Stack-SPOT-Trial-Reward-Masked-Testing
Random Testing results:
{'trial_success_rate_best_index': 992, 'place_success_rate_best_value': 0.836027713625866, 'action_efficiency_best_value': 0.5987903225806451, 'trial_success_rate_best_value': 0.98, 'place_success_rate_best_index': 994, 'grasp_success_rate_best_value': 0.775, 'action_efficiency_best_index': 994, 'grasp_success_rate_best_index': 992}
Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-15-09-59-06_Sim-Stack-SPOT-Trial-Reward-Masked-Training
Training results:
{'trial_success_rate_best_index': 10885, 'place_success_rate_best_value': 0.9295154185022027, 'action_efficiency_best_value': 0.84, 'trial_success_rate_best_value': 0.9166666666666666, 'place_success_rate_best_index': 11876, 'grasp_success_rate_best_index': 13708, 'action_efficiency_best_index': 13536, 'grasp_success_rate_best_value': 0.9416342412451362}
SIM STACK - "SITUATION REMOVAL" - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-22
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 19965 --future_reward_discount 0.9 --max_train_actions 20000 --random_actions --no_height_reward
BAD RUN RESUME: ± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 19965 --future_reward_discount 0.9 --max_train_actions 20000 --random_actions --no_height_reward --resume /home/ahundt/src/real_good_robot/logs/2020-05-18-20-35-14_Sim-Stack-Two-Step-Reward-Training
BAD RUN, used 0.9 discount rather than 0.65: Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-18-20-35-14_Sim-Stack-Two-Step-Reward-Training
Creating data logging session: '/home/ahundt/src/real_good_robot/logs/2020-05-22-14-57-54_Sim-Stack-Two-Step-Reward-Training'
OK RESUME: ± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 19965 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward --resume '/home/ahundt/src/real_good_robot/logs/2020-05-22-14-57-54_Sim-Stack-Two-Step-Reward-Training'
Commit: 7cbb47979cdb12b2e1125fd777a6617b0d5192f9
GPU 0, Tab 0, port 19965, left v-rep window, v-rep tab 7
> Trial logging complete: 101 --------------------------------------------------------------
> Running two step backprop()
> Primitive confidence scores: 1.104833 (push), 1.966652 (grasp), 2.310889 (place)
> Action: grasp at (15, 23, 119)
> Training loss: 0.027554
> prev_height: 0.0 max_z: 0.10344188583902121 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> prev_height: 1.0 max_z: 2.068837716780424 goal_success: True needed to reset: False max_workspace_height: 0.6 <<<<<<<<<<<
> check_stack() stack_height: 2.068837716780424 stack matches current goal: True partial_stack_success: True Does the code think a reset is needed: False
> STACK: trial: 101 actions/partial: 4.230636833046471 actions/full stack: 25.604166666666668 (lower is better) Grasp Count: 1337, grasp success rate: 0.8145100972326104 place_on_stack_rate: 0.5340073529411765 place_attempts: 1088 partial_stack_successes: 581 stack_successes: 96 trial_success_rate: 0.9504950495049505 stack goal: None current_height: 2.068837716780424
> trial_complete_indices: [ 15. 26. 33. 50. 61. 67. 79. 87. 119. 159. 168. 176.
> 191. 205. 209. 248. 258. 267. 291. 303. 330. 371. 397. 411.
> 418. 434. 446. 454. 463. 470. 478. 484. 495. 505. 519. 527.
> 548. 685. 689. 704. 747. 755. 769. 830. 837. 846. 852. 862.
> 873. 880. 886. 909. 915. 921. 931. 1158. 1202. 1215. 1403. 1495.
> 1509. 1520. 1531. 1553. 1565. 1573. 1581. 1592. 1628. 1634. 1640. 1646.
> 1734. 1741. 1749. 1761. 1767. 1773. 1780. 1792. 1804. 1816. 1850. 1862.
> 1882. 2240. 2272. 2289. 2321. 2331. 2335. 2341. 2352. 2358. 2386. 2392.
> 2398. 2404. 2419. 2425. 2457.]
> Max trial success rate: 0.95, at action iteration: 2454. (total of 2456 actions, max excludes first 2454 actions)
> Max grasp success rate: 0.8149812734082397, at action iteration: 2454. (total of 2456 actions, max excludes first 2454 actions)
> Max place success rate: 0.6660714285714285, at action iteration: 2454. (total of 2457 actions, max excludes first 2454 actions)
> Max action efficiency: 0.23471882640586797, at action iteration: 2456. (total of 2457 actions, max excludes first 2454 actions)
> saving plot: 2020-05-28-12-51-52_Sim-Stack-Two-Step-Reward-Testing-Sim-Stack-Two-Step-Reward-Testing_success_plot.png
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-28-12-51-52_Sim-Stack-Two-Step-Reward-Testing/data/best_stats.json
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-28-12-51-52_Sim-Stack-Two-Step-Reward-Testing/best_stats.json
> Choosing a snapshot from the following options:{'grasp_success_rate_best_value': 0.953125, 'grasp_success_rate_best_index': 9853, 'place_success_rate_best_value': 0.7723214285714286, 'place_success_rate_best_index': 19470, 'action_efficiency_best_index': 15291, 'trial_success_rate_best_index': 19128, 'trial_success_rate_best_value': 0.6888888888888889, 'action_efficiency_best_value': 0.468}
> Evaluating trial_success_rate_best_value
> Shapshot chosen: /home/ahundt/src/real_good_robot/logs/2020-05-22-14-57-54_Sim-Stack-Two-Step-Reward-Training/models/snapshot.reinforcement_trial_success_rate_best_value.pth
> Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-22-14-57-54_Sim-Stack-Two-Step-Reward-Training/2020-05-28-12-51-52_Sim-Stack-Two-Step-Reward-Testing
> Random Testing results:
> {'grasp_success_rate_best_value': 0.8149812734082397, 'grasp_success_rate_best_index': 2454, 'place_success_rate_best_value': 0.6660714285714285, 'trial_success_rate_best_index': 2454, 'action_efficiency_best_index': 2456, 'place_success_rate_best_index': 2454, 'trial_success_rate_best_value': 0.95, 'action_efficiency_best_value': 0.23471882640586797}
> Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-22-14-57-54_Sim-Stack-Two-Step-Reward-Training
> Training results:
> {'grasp_success_rate_best_value': 0.953125, 'grasp_success_rate_best_index': 9853, 'place_success_rate_best_value': 0.7723214285714286, 'place_success_rate_best_index': 19470, 'action_efficiency_best_index': 15291, 'trial_success_rate_best_index': 19128, 'trial_success_rate_best_value': 0.6888888888888889, 'action_efficiency_best_value': 0.468}
SIM ROW - "SITUATION REMOVAL" - Mixed RANDOM ACTION, 2D ACTION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-18
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --check_row --place --tcp_port 19999 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-18-20-27-01_Sim-Rows-Two-Step-Reward-Training
Commit: 7cbb47979cdb12b2e1125fd777a6617b0d5192f9
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
> Trial logging complete: 101 --------------------------------------------------------------
> Running two step backprop()
> Primitive confidence scores: 1.917995 (push), 2.243101 (grasp), 2.620736 (place)
> Action: grasp at (11, 199, 104)
> Training loss: 0.002939
> Executing: grasp at (-0.516000, 0.174000, 0.001000) orientation: 4.319690
> gripper position: 0.030604511499404907
> gripper position: 0.026671990752220154
> gripper position: 0.0016168132424354553
> gripper position: -0.022851087152957916
> gripper position: -0.04249643534421921
> Grasp successful: False
> prev_height: 0.0 max_z: 0.05116439476365388 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> check_row: True | row_size: 2 | blocks: ['blue' 'red']
> check_stack() stack_height: 2 stack matches current goal: True partial_stack_success: True Does the code think a reset is needed: False
> STACK: trial: 101 actions/partial: 15.429718875502008 actions/full stack: 42.68888888888889 (lower is better) Grasp Count: 2064, grasp success rate: 0.8507751937984496 place_on_stack_rate: 0.14212
> 32876712329 place_attempts: 1752 partial_stack_successes: 249 stack_successes: 90 trial_success_rate: 0.8910891089108911 stack goal: [1 0] current_height: 2
> trial_complete_indices: [ 6. 18. 36. 81. 158. 164. 330. 1412. 1422. 1432. 1438. 1440.
> 1452. 1534. 1564. 1582. 1588. 1639. 1651. 1667. 1676. 1677. 1696. 1700.
> 1718. 1720. 1726. 1749. 1753. 1757. 1836. 1838. 1845. 1951. 1959. 1961.
> 2058. 2072. 2106. 2108. 2117. 2122. 2160. 2219. 2235. 2239. 2253. 2261.
> 2317. 2323. 2336. 2344. 2363. 2366. 2389. 2391. 2459. 2463. 2473. 2691.
> 2698. 2711. 2717. 2734. 2766. 2770. 2772. 2780. 2790. 2792. 2794. 2813.
> 2822. 2830. 2841. 2851. 2886. 2915. 2988. 3201. 3231. 3233. 3239. 3257.
> 3281. 3302. 3435. 3453. 3468. 3485. 3493. 3521. 3525. 3631. 3679. 3697.
> 3707. 3720. 3820. 3834. 3841.]
> Max trial success rate: 0.89, at action iteration: 3838. (total of 3840 actions, max excludes first 3838 actions)
> Max grasp success rate: 0.8516003879728419, at action iteration: 3839. (total of 3840 actions, max excludes first 3838 actions)
> Max place success rate: 0.6004566210045662, at action iteration: 3840. (total of 3841 actions, max excludes first 3838 actions)
> Max action efficiency: 0.1422615945805107, at action iteration: 3840. (total of 3841 actions, max excludes first 3838 actions)
> saving plot: 2020-05-24-16-22-22_Sim-Rows-Two-Step-Reward-Testing-Sim-Rows-Two-Step-Reward-Testing_success_plot.png
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-24-16-22-22_Sim-Rows-Two-Step-Reward-Testing/data/best_stats.json
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-05-24-16-22-22_Sim-Rows-Two-Step-Reward-Testing/best_stats.json
> TODO(ahundt) if there is time, try the action efficiency version
> Choosing a snapshot from the following options:{'place_success_rate_best_value': 0.7122641509433962, 'action_efficiency_best_value': 0.588, 'grasp_success_rate_best_index': 19749, 'place_success_rate
> _best_index': 14616, 'trial_success_rate_best_index': 13203, 'grasp_success_rate_best_value': 0.7829181494661922, 'action_efficiency_best_index': 14623, 'trial_success_rate_best_value': 0.5138888888888888}
> Evaluating trial_success_rate_best_value
> The trial_success_rate_best_value is fantastic at 0.5138888888888888, so we will look for the best action_efficiency_best_value.
> Shapshot chosen: /home/ahundt/src/real_good_robot/logs/2020-05-18-20-27-01_Sim-Rows-Two-Step-Reward-Training/models/snapshot.reinforcement_action_efficiency_best_value.pth
> testing snapshot, prioritizing action efficiency: /home/ahundt/src/real_good_robot/logs/2020-05-18-20-27-01_Sim-Rows-Two-Step-Reward-Training/models/snapshot.reinforcement_trial_success_rate_best_value.pth
SIM STACK - "Baseline" - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-22
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward --disable_situation_removal
Resume: ± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward --disable_situation_removal --resume /home/ahundt/src/real_good_robot/logs/2020-05-22-20-49-56_Sim-Stack-Two-Step-Reward-Training/
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-22-20-49-56_Sim-Stack-Two-Step-Reward-Training/
Commit: 2d5e56817c0d20120b31261778bdfc1011b1d623 same as tag v0.16.0
GPU 2, Tab 2, port 20000, right v-rep window, v-rep tab 9
> TODO(ahundt) consider trying an extra test run
> note "best stats" look better than reality because about 2 trials ended in a 500 action span.
> Training "best stats" {"action_efficiency_best_index": 15291, "action_efficiency_best_value": 0.468, "grasp_success_rate_best_index": 9853, "grasp_success_rate_best_value": 0.953125, "place_success_rate_best_index": 19470, "place_success_rate_best_value": 0.7723214285714286, "trial_success_rate_best_index": 19128, "trial_success_rate_best_value": 0.6888888888888889}
> % /home/ahundt/src/real_good_robot/logs/2020-05-22-20-49-56_Sim-Stack-Two-Step-Reward-Training/ only 2 trials completed successfully during 20k actions of training with a max training efficiency of 1%.
> Training iteration: 20006
> prev_height: 0.0 max_z: 0.05113055666990403 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> Current count of pixels with stuff: 4531.0 threshold below which the scene is considered empty: 300
> WARNING variable mismatch num_trials + 1: 128 nonlocal_variables[stack].trial: 126
> Change detected: True (value: 534)
> Primitive confidence scores: 1.406690 (push), 2.592584 (grasp), 2.803117 (place)
> Strategy: exploit (exploration probability: 0.010000)
> Action: place at (0, 104, 103)
> Executing: Place at (-0.518000, -0.016000, 0.051012) angle: 0.000000
> gripper position: 0.005653828382492065
> gripper position: 0.0055609047412872314
> gripper position: 0.005279242992401123
> Trainer.get_label_value(): Current reward: 1.000000 Current reward multiplier: 1.000000 Predicted Future reward: 2.803117 Expected reward: 1.000000 + 0.650000 x 2.803117 = 2.822026
> Running two step backprop()
> Training loss: 0.004022
> current_position: [-0.52026516 -0.01241153 0.07798614]
> current_obj_z_location: 0.10798614352941513
> goal_position: 0.11101188566865967 goal_position_margin: 0.31101188566865967
> has_moved: True near_goal: True place_success: True
> prev_height: 0.0 max_z: 0.1031934318821809 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> prev_height: 1.0232359626064527 max_z: 2.063868637643618 goal_success: True needed to reset: False max_workspace_height: 0.6232359626064526 <<<<<<<<<<<
> check_stack() stack_height: 2.063868637643618 stack matches current goal: True partial_stack_success: True Does the code think a reset is needed: False
> STACK: trial: 126 actions/partial: 2.941773268636965 actions/full stack: 20007.0 (lower is better) Grasp Count: 10612, grasp success rate: 0.8646814926498304 place_on_stack_rate: 0.7427916120576671 place_
> attempts: 9156 partial_stack_successes: 6801 stack_successes: 1 trial_success_rate: 0.007936507936507936 stack goal: None current_height: 2.063868637643618
> Experience replay 18488: history timestep index 205, action: push, surprise value: 1.354873
> Training loss: 0.000761
> Time elapsed: 22.195343
> Trainer iteration: 20006 complete
SIM ROW - "Baseline" - Mixed RANDOM ACTION, 2D ACTION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-25
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --check_row --place --tcp_port 19999 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward --disable_situation_removal
RESUME: export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --check_row --place --tcp_port 19999 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward --disable_situation_removal --resume /home/ahundt/src/real_good_robot/logs/2020-05-25-14-13-02_Sim-Rows-Two-Step-Reward-Training
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-25-14-13-02_Sim-Rows-Two-Step-Reward-Training
Commit: a534735959ec2747c3b134a6d3067135a5c7bd75 same as tag v0.16.0
GPU 1, Tab 1, port 19999, middle v-rep window, v-rep tab 8
> First trial took 1200 actions, manually resetting to the next trial didn't improve, so rounding down to 0.
> Testing iteration: 1206
> prev_height: 0.0 max_z: 0.051105897527620084 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> Current count of pixels with stuff: 1804.0 threshold below which the scene is considered empty: 900
> Change detected: True (value: 523)
> Primitive confidence scores: 1.369749 (push), 3.314228 (grasp), 2.010027 (place)
> Action: place at (12, 39, 215)
> Executing: Place at (-0.294000, -0.146000, 0.001007) angle: 4.712389
> gripper position: 0.0044525861740112305
> gripper position: 0.0043984055519104
> gripper position: 0.0042927563190460205
> Trainer.get_label_value(): Current reward: 1.000000 Current reward multiplier: 1.000000 Predicted Future reward: 3.314228 Expected reward: 1.000000 + 0.650000 x 3.314228 = 3.154248
> Running two step backprop()
> Training loss: 0.012545
> current_position: [-0.2937963 -0.1476554 0.0259938]
> current_obj_z_location: 0.05599379979074001
> goal_position: 0.021007411514566848 goal_position_margin: 0.22100741151456685
> has_moved: True near_goal: True place_success: True
> prev_height: 0.0 max_z: 0.0511212507029354 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> check_row: True | row_size: 2 | blocks: ['blue' 'green']
> check_stack() stack_height: 2 stack matches current goal: False partial_stack_success: False Does the code think a reset is needed: True
> main.py check_stack() DETECTED PROGRESS REVERSAL, mismatch between the goal height: 3 and current workspace stack height: 2
> STACK: trial: 1 actions/partial: 402.3333333333333 actions/full stack: inf (lower is better) Grasp Count: 605, grasp success rate: 0.996694214876033 place_on_stack_rate: 0.0049833887043189366 place_attemp
> ts: 602 partial_stack_successes: 3 stack_successes: 0 trial_success_rate: inf stack goal: [0 2 1 3] current_height: 2
> Time elapsed: 15.938629
XXX do not use - SIM STACK - "SITUATION REMOVAL" - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-29
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 19965 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-29-11-03-34_Sim-Stack-Two-Step-Reward-Training
Commit: f8c4d93db4a48b905b8995188de357e92171ae10
GPU 0, Tab 0, port 19965, left v-rep window, v-rep tab 7
We had hard disk io problems on this run, leading to many resets during training.
> Trial logging complete: 101 --------------------------------------------------------------
> Running two step backprop()
> Primitive confidence scores: 1.256993 (push), 2.075450 (grasp), 1.718884 (place)
> Action: grasp at (4, 24, 199)
> Training loss: 0.485695
> Executing: grasp at (-0.326000, -0.176000, 0.001006) orientation: 1.570796
> gripper position: 0.03197145462036133
> gripper position: 0.026409685611724854
> gripper position: 0.0013526976108551025
> gripper position: -0.023178979754447937
> gripper position: -0.041759684681892395
> Grasp successful: False
> prev_height: 0.0 max_z: 0.051128429751611595 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> prev_height: 1.0 max_z: 1.0225685950322319 goal_success: False needed to reset: False max_workspace_height: 0.6 <<<<<<<<<<<
> check_stack() stack_height: 1.0225685950322319 stack matches current goal: False partial_stack_success: False Does the code think a reset is needed: False
> STACK: trial: 101 actions/partial: 6.061032863849765 actions/full stack: 40.34375 (lower is better) Grasp Count: 758, grasp success rate: 0.7031662269129287 place_on_stack_rate: 0.399624765478424 place_attempts: 533 partial_stack_successes: 213 stack_successes: 32 trial_success_rate: 0.31683168316831684 stack goal: None current_height: 1.0225685950322319
> trial_complete_indices: [ 45. 130. 134. 164. 200. 207. 215. 252. 260. 261. 262. 284.
> 302. 309. 313. 323. 339. 356. 392. 396. 397. 398. 399. 400.
> 431. 447. 448. 457. 473. 481. 499. 507. 515. 528. 536. 552.
> 553. 555. 557. 566. 580. 609. 638. 686. 688. 689. 690. 697.
> 698. 699. 719. 761. 826. 828. 841. 845. 846. 847. 848. 849.
> 850. 851. 852. 853. 854. 855. 856. 857. 858. 859. 860. 861.
> 862. 863. 866. 870. 896. 909. 910. 927. 964. 974. 977. 1001.
> 1024. 1026. 1048. 1057. 1064. 1074. 1082. 1088. 1102. 1114. 1120. 1145.
> 1163. 1213. 1222. 1272. 1290.]
> Max trial success rate: 0.32, at action iteration: 1287. (total of 1289 actions, max excludes first 1287 actions)
> Max grasp success rate: 0.7037037037037037, at action iteration: 1287. (total of 1289 actions, max excludes first 1287 actions)
> Max place success rate: 0.6842105263157895, at action iteration: 1287. (total of 1290 actions, max excludes first 1287 actions)
> Max action efficiency: 0.1351981351981352, at action iteration: 1287. (total of 1290 actions, max excludes first 1287 actions)
> saving trial success rate: /home/ahundt/src/real_good_robot/logs/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing/transitions/trial-success-rate.log.csv
> saving grasp success rate: /home/ahundt/src/real_good_robot/logs/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing/transitions/grasp-success-rate.log.csv
> saving place success rate: /home/ahundt/src/real_good_robot/logs/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing/transitions/place-success-rate.log.csv
> saving action efficiency: /home/ahundt/src/real_good_robot/logs/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing/transitions/action-efficiency.log.csv
> saving plot: 2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing-Sim-Stack-Two-Step-Reward-Testing_success_plot.png
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing/data/best_stats.json
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing/best_stats.json
> Choosing a snapshot from the following options:{'trial_success_rate_best_value': 0.5454545454545454, 'action_efficiency_best_value': 0.324, 'grasp_success_rate_best_index': 17743, 'place_success_rate_best_index': 15811, 'place_success_rate_best_value': 0.7477064220183486, 'action_efficiency_best_index': 9542, 'grasp_success_rate_best_value': 0.8901515151515151, 'trial_success_rate_best_index': 15666}
> Evaluating trial_success_rate_best_value
> Shapshot chosen: /home/ahundt/src/real_good_robot/logs/2020-05-29-11-03-34_Sim-Stack-Two-Step-Reward-Training/models/snapshot.reinforcement_trial_success_rate_best_value.pth
> Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-29-11-03-34_Sim-Stack-Two-Step-Reward-Training/2020-06-07-13-32-32_Sim-Stack-Two-Step-Reward-Testing
> Random Testing results:
> {'trial_success_rate_best_value': 0.32, 'action_efficiency_best_value': 0.1351981351981352, 'grasp_success_rate_best_index': 1287, 'place_success_rate_best_index': 1287, 'place_success_rate_best_value': 0.6842105263157895, 'action_efficiency_best_index': 1287, 'grasp_success_rate_best_value': 0.7037037037037037, 'trial_success_rate_best_index': 1287}
> Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-29-11-03-34_Sim-Stack-Two-Step-Reward-Training
> Training results:
> {'trial_success_rate_best_value': 0.5454545454545454, 'action_efficiency_best_value': 0.324, 'grasp_success_rate_best_index': 17743, 'place_success_rate_best_index': 15811, 'place_success_rate_best_value': 0.7477064220183486, 'action_efficiency_best_index': 9542, 'grasp_success_rate_best_value': 0.8901515151515151, 'trial_success_rate_best_index': 15666}
± '/home/ahundt/src/real_good_robot/logs/2020-06-08-16-05-39_Sim-Stack-Two-Step-Reward-Testing/best_stats.json'
{"action_efficiency_best_index": 8960, "action_efficiency_best_value": 0.04018754186202277, "grasp_success_rate_best_index": 8959, "grasp_success_rate_best_value": 0.8608226007478189, "place_success_rate_best_index": 8960, "place_success_rate_best_value": 0.7477086348287506, "trial_success_rate_best_index": 8958, "trial_success_rate_best_value": 0.8405797101449275}
SIM ROW - "SITUATION REMOVAL" - Mixed RANDOM ACTION, 2D ACTION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05- TODO
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --check_row --place --tcp_port 19999 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --no_height_reward
Creating data logging session: TODO
Commit: 12d9481717486342dbfcaff191ddb1428f102406
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
SIM STACK - "instant Task Progress aka progress only aka Rp" - Mixed RANDOM ACTION, 2D ACTION EXPLORATION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-30
---------------------------------------------------------------------------------------------
± export CUDA_VISIBLE_DEVICES="2" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 20000 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions
Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-30-22-11-28_Sim-Stack-Two-Step-Reward-Training
Commit: 12d9481717486342dbfcaff191ddb1428f102406
GPU 2, Tab 2, port 20000, left v-rep window, v-rep tab 7
RESUME ON GPU 0: ± export CUDA_VISIBLE_DEVICES="0" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 8 --push_rewards --experience_replay --explore_rate_decay --check_z_height --place --tcp_port 19965 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions --resume /home/ahundt/src/real_good_robot/logs/2020-05-30-22-11-28_Sim-Stack-Two-Step-Reward-Training/
XXX do not use - SIM ROW - "instant Task Progress aka progress only aka Rp" - Mixed RANDOM ACTION, 2D ACTION - REWARD SCHEDULE 0.1, 1, 1 - femur 2020-05-30
-------------------------------------------------------------------------------------------
export CUDA_VISIBLE_DEVICES="1" && python3 main.py --is_sim --obj_mesh_dir objects/blocks --num_obj 4 --push_rewards --experience_replay --explore_rate_decay --check_row --place --tcp_port 19999 --future_reward_discount 0.65 --max_train_actions 20000 --random_actions
COMPUTER WENT DOWN: Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-05-30-21-42-56_Sim-Rows-Two-Step-Reward-Training
Followup run: Creating data logging session: /home/ahundt/src/real_good_robot/logs/2020-06-04-14-03-44_Sim-Stack-Two-Step-Reward-Training
Commit: 12d9481717486342dbfcaff191ddb1428f102406
GPU 1, Tab 1, port 19999, right v-rep window, v-rep tab 8
there were disk io problems during this training run, leading to many resets & lost data.
> Trial logging complete: 101 --------------------------------------------------------------
> prev_height: 0.0 max_z: 0.05112180110319438 goal_success: True needed to reset: False max_workspace_height: -0.02 <<<<<<<<<<<
> check_row: True | row_size: 2 | blocks: ['green' 'yellow']
> check_stack() stack_height: 2 stack matches current goal: True partial_stack_success: True Does the code think a reset is needed: False
> STACK: trial: 101 actions/partial: 5.116161616161616 actions/full stack: 10.552083333333334 (lower is better) Grasp Count: 589, grasp success rate: 0.7198641765704584 place_on_stack_rate: 0.4669811320754717 place_attempts: 424 partial_stack_successes: 198 stack_successes: 96 trial_success_rate: 0.9504950495049505 stack goal: [1 2] current_height: 2
> trial_complete_indices: [ 3. 8. 12. 19. 24. 28. 33. 47. 52. 64. 70. 74.
> 224. 231. 237. 243. 247. 249. 259. 267. 271. 278. 290. 302.
> 307. 310. 315. 320. 322. 326. 331. 334. 336. 343. 347. 349.
> 356. 358. 373. 376. 384. 396. 406. 525. 531. 556. 560. 569.
> 575. 577. 583. 589. 603. 609. 621. 642. 649. 653. 658. 665.
> 671. 673. 677. 688. 697. 706. 712. 753. 755. 759. 792. 797.
> 801. 805. 807. 809. 813. 825. 829. 852. 856. 864. 868. 872.
> 882. 895. 900. 907. 911. 913. 922. 927. 934. 947. 959. 963.
> 989. 993. 1003. 1007. 1012.]
> Max trial success rate: 0.95, at action iteration: 1009. (total of 1011 actions, max excludes first 1009 actions)
> Max grasp success rate: 0.7223168654173765, at action iteration: 1010. (total of 1011 actions, max excludes first 1009 actions)
> Max place success rate: 0.8632075471698113, at action iteration: 1011. (total of 1012 actions, max excludes first 1009 actions)
> Max action efficiency: 0.576808721506442, at action iteration: 1011. (total of 1012 actions, max excludes first 1009 actions)
> saving trial success rate: /home/ahundt/src/real_good_robot/logs/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing/transitions/trial-success-rate.log.csv
> saving grasp success rate: /home/ahundt/src/real_good_robot/logs/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing/transitions/grasp-success-rate.log.csv
> saving place success rate: /home/ahundt/src/real_good_robot/logs/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing/transitions/place-success-rate.log.csv
> saving action efficiency: /home/ahundt/src/real_good_robot/logs/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing/transitions/action-efficiency.log.csv
> saving plot: 2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing-Sim-Rows-Two-Step-Reward-Testing_success_plot.png
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing/data/best_stats.json
> saving best stats to: /home/ahundt/src/real_good_robot/logs/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing/best_stats.json
> Choosing a snapshot from the following options:{'grasp_success_rate_best_value': 0.8122743682310469, 'action_efficiency_best_index': 19045, 'grasp_success_rate_best_index': 15802, 'action_efficiency_best_value': 1.104, 'place_success_rate_best_index': 18289, 'trial_success_rate_best_index': 19892, 'place_success_rate_best_value': 0.8916256157635468, 'trial_success_rate_best_value': 0.7033898305084746}
> Evaluating trial_success_rate_best_value
> The trial_success_rate_best_value is fantastic at 0.7033898305084746, so we will look for the best action_efficiency_best_value.
> Shapshot chosen: /home/ahundt/src/real_good_robot/logs/2020-05-30-21-42-56_Sim-Rows-Two-Step-Reward-Training/models/snapshot.reinforcement_action_efficiency_best_value.pth
> Random Testing Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-30-21-42-56_Sim-Rows-Two-Step-Reward-Training/2020-06-09-19-33-43_Sim-Rows-Two-Step-Reward-Testing
> Random Testing results:
> {'grasp_success_rate_best_value': 0.7223168654173765, 'action_efficiency_best_index': 1011, 'grasp_success_rate_best_index': 1010, 'action_efficiency_best_value': 0.576808721506442, 'place_success_rate_best_index': 1011, 'trial_success_rate_best_index': 1009, 'place_success_rate_best_value': 0.8632075471698113, 'trial_success_rate_best_value': 0.95}
> Training Complete! Dir: /home/ahundt/src/real_good_robot/logs/2020-05-30-21-42-56_Sim-Rows-Two-Step-Reward-Training
> Training results:
> {'grasp_success_rate_best_value': 0.8122743682310469, 'action_efficiency_best_index': 19045, 'grasp_success_rate_best_index': 15802, 'action_efficiency_best_value': 1.104, 'place_success_rate_best_index': 18289, 'trial_success_rate_best_index': 19892, 'place_success_rate_best_value': 0.8916256157635468, 'trial_success_rate_best_value': 0.7033898305084746}
TODO:
"No Reversal" (maybe rename basic progress)
=============================================================
=============================================================