Hugging Face's logo

Datasets:
MLL-Lab
/
LAGEN-datasets

Tasks:
Robotics
Modalities:
Image
Tabular
Text
Size:
10M - 100M
Tags:
LeRobot
image
video
License:
Dataset card Data Studio Files
xet
 Community
1
Caesarrr commited on
Commit
0f1ebf0
·
verified ·
1 Parent(s): fb88acf

Consolidate task-transfer assets 1-80

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/.gitattributes +35 -0
  2. archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_1000.json +531 -0
  3. archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_1500.json +532 -0
  4. archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_2000.json +535 -0
  5. archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_500.json +535 -0
  6. archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/post_train/step_2000.json +1177 -0
  7. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/.gitattributes +37 -0
  8. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/_initialization/StarVLA_Qwen3VL-OFT-Bridge-RT-1/checkpoints/steps_5000_pytorch_model.pt +3 -0
  9. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/steps_15000_state/optimizer.bin +3 -0
  10. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/steps_15000_state/random_states_0.pkl +3 -0
  11. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/eval/mid_train/step_1000.json +63 -0
  12. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/eval/post_train/step_15000.json +993 -0
  13. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/hydra/train_starvla_hydra.log +0 -0
  14. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/summary.jsonl +1 -0
  15. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/debug-internal.log +0 -0
  16. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/debug.log +25 -0
  17. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/files/wandb-summary.json +1 -0
  18. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug-core.log +14 -0
  19. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug-internal.log +17 -0
  20. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug.log +13 -0
  21. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/run-r8358sud.wandb +0 -0
  22. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/config.yaml +417 -0
  23. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/output.log +0 -0
  24. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/requirements.txt +227 -0
  25. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/wandb-metadata.json +387 -0
  26. archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/wandb-summary.json +1 -0
  27. experiments/task-transfer/provenance/asterix_200ep/.gitattributes +60 -0
  28. experiments/task-transfer/provenance/asterix_200ep/README.md +133 -0
  29. experiments/task-transfer/provenance/deadly_1000ep/.gitattributes +60 -0
  30. experiments/task-transfer/provenance/defend_the_line_200ep/.gitattributes +60 -0
  31. experiments/task-transfer/provenance/defend_the_line_200ep/README.md +127 -0
  32. latency-aware/asterix/asterix_200ep/asterix_fixed_latency_2_200ep_7k2steps/train-00000-of-00001.parquet +3 -0
  33. latency-aware/asterix/asterix_200ep/asterix_fixed_latency_2_200ep_7k2steps/val-00000-of-00001.parquet +3 -0
  34. latency-aware/asterix/asterix_200ep/asterix_fixed_latency_4_200ep_7k2steps/train-00000-of-00001.parquet +3 -0
  35. latency-aware/asterix/asterix_200ep/asterix_fixed_latency_4_200ep_7k2steps/val-00000-of-00001.parquet +3 -0
  36. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/filter_report.json +12 -0
  37. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/metadata.json +350 -0
  38. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/train.parquet +3 -0
  39. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/val.parquet +3 -0
  40. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/filter_report.json +12 -0
  41. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/metadata.json +350 -0
  42. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/probe_state.json +12 -0
  43. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/train.parquet +3 -0
  44. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/val.parquet +3 -0
  45. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/filter_report.json +12 -0
  46. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/metadata.json +350 -0
  47. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/train.parquet +3 -0
  48. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/val.parquet +3 -0
  49. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_8_1000ep/filter_report.json +12 -0
  50. latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_8_1000ep/metadata.json +350 -0
archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_1000.json ADDED
@@ -0,0 +1,531 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 5,
6
+ "mean_reward": -9.299999999999999,
7
+ "mean_length": 50.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ -9.299999999999999,
12
+ -9.299999999999999,
13
+ -9.299999999999999,
14
+ -9.299999999999999,
15
+ -9.299999999999999
16
+ ],
17
+ "episode_lengths": [
18
+ 50,
19
+ 50,
20
+ 50,
21
+ 50,
22
+ 50
23
+ ],
24
+ "decoded_action_hist": {
25
+ "1": 250
26
+ },
27
+ "fixed_episode_seeds": true,
28
+ "eval_seed": 42,
29
+ "episode_seeds": [
30
+ 42,
31
+ 43,
32
+ 44,
33
+ 45,
34
+ 46
35
+ ],
36
+ "episode_indices": [
37
+ 0,
38
+ 1,
39
+ 2,
40
+ 3,
41
+ 4
42
+ ]
43
+ },
44
+ "flappy/latency_1": {
45
+ "latency": 1,
46
+ "num_episodes": 5,
47
+ "mean_reward": -9.299999999999999,
48
+ "mean_length": 50.0,
49
+ "std_reward": 0.0,
50
+ "std_length": 0.0,
51
+ "episode_rewards": [
52
+ -9.299999999999999,
53
+ -9.299999999999999,
54
+ -9.299999999999999,
55
+ -9.299999999999999,
56
+ -9.299999999999999
57
+ ],
58
+ "episode_lengths": [
59
+ 50,
60
+ 50,
61
+ 50,
62
+ 50,
63
+ 50
64
+ ],
65
+ "decoded_action_hist": {
66
+ "0": 5,
67
+ "1": 245
68
+ },
69
+ "fixed_episode_seeds": true,
70
+ "eval_seed": 42,
71
+ "episode_seeds": [
72
+ 42,
73
+ 43,
74
+ 44,
75
+ 45,
76
+ 46
77
+ ],
78
+ "episode_indices": [
79
+ 0,
80
+ 1,
81
+ 2,
82
+ 3,
83
+ 4
84
+ ]
85
+ },
86
+ "flappy/latency_2": {
87
+ "latency": 2,
88
+ "num_episodes": 5,
89
+ "mean_reward": -9.299999999999999,
90
+ "mean_length": 50.0,
91
+ "std_reward": 0.0,
92
+ "std_length": 0.0,
93
+ "episode_rewards": [
94
+ -9.299999999999999,
95
+ -9.299999999999999,
96
+ -9.299999999999999,
97
+ -9.299999999999999,
98
+ -9.299999999999999
99
+ ],
100
+ "episode_lengths": [
101
+ 50,
102
+ 50,
103
+ 50,
104
+ 50,
105
+ 50
106
+ ],
107
+ "decoded_action_hist": {
108
+ "0": 10,
109
+ "1": 240
110
+ },
111
+ "fixed_episode_seeds": true,
112
+ "eval_seed": 42,
113
+ "episode_seeds": [
114
+ 42,
115
+ 43,
116
+ 44,
117
+ 45,
118
+ 46
119
+ ],
120
+ "episode_indices": [
121
+ 0,
122
+ 1,
123
+ 2,
124
+ 3,
125
+ 4
126
+ ]
127
+ },
128
+ "flappy/latency_3": {
129
+ "latency": 3,
130
+ "num_episodes": 5,
131
+ "mean_reward": -9.299999999999999,
132
+ "mean_length": 50.0,
133
+ "std_reward": 0.0,
134
+ "std_length": 0.0,
135
+ "episode_rewards": [
136
+ -9.299999999999999,
137
+ -9.299999999999999,
138
+ -9.299999999999999,
139
+ -9.299999999999999,
140
+ -9.299999999999999
141
+ ],
142
+ "episode_lengths": [
143
+ 50,
144
+ 50,
145
+ 50,
146
+ 50,
147
+ 50
148
+ ],
149
+ "decoded_action_hist": {
150
+ "0": 15,
151
+ "1": 235
152
+ },
153
+ "fixed_episode_seeds": true,
154
+ "eval_seed": 42,
155
+ "episode_seeds": [
156
+ 42,
157
+ 43,
158
+ 44,
159
+ 45,
160
+ 46
161
+ ],
162
+ "episode_indices": [
163
+ 0,
164
+ 1,
165
+ 2,
166
+ 3,
167
+ 4
168
+ ]
169
+ },
170
+ "flappy/latency_4": {
171
+ "latency": 4,
172
+ "num_episodes": 5,
173
+ "mean_reward": -8.7,
174
+ "mean_length": 50.0,
175
+ "std_reward": 0.0,
176
+ "std_length": 0.0,
177
+ "episode_rewards": [
178
+ -8.7,
179
+ -8.7,
180
+ -8.7,
181
+ -8.7,
182
+ -8.7
183
+ ],
184
+ "episode_lengths": [
185
+ 50,
186
+ 50,
187
+ 50,
188
+ 50,
189
+ 50
190
+ ],
191
+ "decoded_action_hist": {
192
+ "0": 20,
193
+ "1": 230
194
+ },
195
+ "fixed_episode_seeds": true,
196
+ "eval_seed": 42,
197
+ "episode_seeds": [
198
+ 42,
199
+ 43,
200
+ 44,
201
+ 45,
202
+ 46
203
+ ],
204
+ "episode_indices": [
205
+ 0,
206
+ 1,
207
+ 2,
208
+ 3,
209
+ 4
210
+ ]
211
+ },
212
+ "flappy/latency_5": {
213
+ "latency": 5,
214
+ "num_episodes": 5,
215
+ "mean_reward": -8.7,
216
+ "mean_length": 50.0,
217
+ "std_reward": 0.0,
218
+ "std_length": 0.0,
219
+ "episode_rewards": [
220
+ -8.7,
221
+ -8.7,
222
+ -8.7,
223
+ -8.7,
224
+ -8.7
225
+ ],
226
+ "episode_lengths": [
227
+ 50,
228
+ 50,
229
+ 50,
230
+ 50,
231
+ 50
232
+ ],
233
+ "decoded_action_hist": {
234
+ "0": 25,
235
+ "1": 225
236
+ },
237
+ "fixed_episode_seeds": true,
238
+ "eval_seed": 42,
239
+ "episode_seeds": [
240
+ 42,
241
+ 43,
242
+ 44,
243
+ 45,
244
+ 46
245
+ ],
246
+ "episode_indices": [
247
+ 0,
248
+ 1,
249
+ 2,
250
+ 3,
251
+ 4
252
+ ]
253
+ },
254
+ "demon_attack/latency_0": {
255
+ "latency": 0,
256
+ "num_episodes": 5,
257
+ "mean_reward": 114.0,
258
+ "mean_length": 3600.0,
259
+ "std_reward": 44.090815370097204,
260
+ "std_length": 0.0,
261
+ "episode_rewards": [
262
+ 150.0,
263
+ 60.0,
264
+ 150.0,
265
+ 150.0,
266
+ 60.0
267
+ ],
268
+ "episode_lengths": [
269
+ 3600,
270
+ 3600,
271
+ 3600,
272
+ 3600,
273
+ 3600
274
+ ],
275
+ "decoded_action_hist": {
276
+ "1": 17017,
277
+ "5": 983
278
+ },
279
+ "fixed_episode_seeds": true,
280
+ "eval_seed": 42,
281
+ "episode_seeds": [
282
+ 42,
283
+ 43,
284
+ 44,
285
+ 45,
286
+ 46
287
+ ],
288
+ "episode_indices": [
289
+ 0,
290
+ 1,
291
+ 2,
292
+ 3,
293
+ 4
294
+ ]
295
+ },
296
+ "demon_attack/latency_1": {
297
+ "latency": 1,
298
+ "num_episodes": 5,
299
+ "mean_reward": 144.0,
300
+ "mean_length": 3600.0,
301
+ "std_reward": 98.8635423197045,
302
+ "std_length": 0.0,
303
+ "episode_rewards": [
304
+ 265.0,
305
+ 70.0,
306
+ 265.0,
307
+ 60.0,
308
+ 60.0
309
+ ],
310
+ "episode_lengths": [
311
+ 3600,
312
+ 3600,
313
+ 3600,
314
+ 3600,
315
+ 3600
316
+ ],
317
+ "decoded_action_hist": {
318
+ "0": 5,
319
+ "1": 16579,
320
+ "3": 8,
321
+ "5": 1408
322
+ },
323
+ "fixed_episode_seeds": true,
324
+ "eval_seed": 42,
325
+ "episode_seeds": [
326
+ 42,
327
+ 43,
328
+ 44,
329
+ 45,
330
+ 46
331
+ ],
332
+ "episode_indices": [
333
+ 0,
334
+ 1,
335
+ 2,
336
+ 3,
337
+ 4
338
+ ]
339
+ },
340
+ "demon_attack/latency_2": {
341
+ "latency": 2,
342
+ "num_episodes": 5,
343
+ "mean_reward": 94.0,
344
+ "mean_length": 3600.0,
345
+ "std_reward": 37.73592452822641,
346
+ "std_length": 0.0,
347
+ "episode_rewards": [
348
+ 140.0,
349
+ 60.0,
350
+ 140.0,
351
+ 60.0,
352
+ 70.0
353
+ ],
354
+ "episode_lengths": [
355
+ 3600,
356
+ 3600,
357
+ 3600,
358
+ 3600,
359
+ 3600
360
+ ],
361
+ "decoded_action_hist": {
362
+ "0": 10,
363
+ "1": 17267,
364
+ "5": 723
365
+ },
366
+ "fixed_episode_seeds": true,
367
+ "eval_seed": 42,
368
+ "episode_seeds": [
369
+ 42,
370
+ 43,
371
+ 44,
372
+ 45,
373
+ 46
374
+ ],
375
+ "episode_indices": [
376
+ 0,
377
+ 1,
378
+ 2,
379
+ 3,
380
+ 4
381
+ ]
382
+ },
383
+ "demon_attack/latency_3": {
384
+ "latency": 3,
385
+ "num_episodes": 5,
386
+ "mean_reward": 153.0,
387
+ "mean_length": 3600.0,
388
+ "std_reward": 62.89674077406555,
389
+ "std_length": 0.0,
390
+ "episode_rewards": [
391
+ 140.0,
392
+ 70.0,
393
+ 140.0,
394
+ 265.0,
395
+ 150.0
396
+ ],
397
+ "episode_lengths": [
398
+ 3600,
399
+ 3600,
400
+ 3600,
401
+ 3600,
402
+ 3600
403
+ ],
404
+ "decoded_action_hist": {
405
+ "0": 15,
406
+ "1": 17589,
407
+ "5": 395,
408
+ "3": 1
409
+ },
410
+ "fixed_episode_seeds": true,
411
+ "eval_seed": 42,
412
+ "episode_seeds": [
413
+ 42,
414
+ 43,
415
+ 44,
416
+ 45,
417
+ 46
418
+ ],
419
+ "episode_indices": [
420
+ 0,
421
+ 1,
422
+ 2,
423
+ 3,
424
+ 4
425
+ ]
426
+ },
427
+ "demon_attack/latency_4": {
428
+ "latency": 4,
429
+ "num_episodes": 5,
430
+ "mean_reward": 136.0,
431
+ "mean_length": 3600.0,
432
+ "std_reward": 93.08061022576076,
433
+ "std_length": 0.0,
434
+ "episode_rewards": [
435
+ 250.0,
436
+ 60.0,
437
+ 250.0,
438
+ 60.0,
439
+ 60.0
440
+ ],
441
+ "episode_lengths": [
442
+ 3600,
443
+ 3600,
444
+ 3600,
445
+ 3600,
446
+ 3600
447
+ ],
448
+ "decoded_action_hist": {
449
+ "0": 20,
450
+ "1": 16921,
451
+ "5": 1059
452
+ },
453
+ "fixed_episode_seeds": true,
454
+ "eval_seed": 42,
455
+ "episode_seeds": [
456
+ 42,
457
+ 43,
458
+ 44,
459
+ 45,
460
+ 46
461
+ ],
462
+ "episode_indices": [
463
+ 0,
464
+ 1,
465
+ 2,
466
+ 3,
467
+ 4
468
+ ]
469
+ },
470
+ "demon_attack/latency_5": {
471
+ "latency": 5,
472
+ "num_episodes": 5,
473
+ "mean_reward": 204.0,
474
+ "mean_length": 1869.8,
475
+ "std_reward": 96.24967532412772,
476
+ "std_length": 1414.8217414218655,
477
+ "episode_rewards": [
478
+ 150.0,
479
+ 250.0,
480
+ 150.0,
481
+ 370.0,
482
+ 100.0
483
+ ],
484
+ "episode_lengths": [
485
+ 787,
486
+ 3600,
487
+ 787,
488
+ 3600,
489
+ 575
490
+ ],
491
+ "decoded_action_hist": {
492
+ "0": 25,
493
+ "1": 8803,
494
+ "5": 521
495
+ },
496
+ "fixed_episode_seeds": true,
497
+ "eval_seed": 42,
498
+ "episode_seeds": [
499
+ 42,
500
+ 43,
501
+ 44,
502
+ 45,
503
+ 46
504
+ ],
505
+ "episode_indices": [
506
+ 0,
507
+ 1,
508
+ 2,
509
+ 3,
510
+ 4
511
+ ]
512
+ }
513
+ },
514
+ "aggregate": {
515
+ "stage": "mid_train",
516
+ "step": 1000,
517
+ "task": "cross_task",
518
+ "model_alias": "openvla",
519
+ "fixed_episode_seeds": true,
520
+ "eval_seed": 42,
521
+ "total_episodes": 60,
522
+ "mean_reward": 65.86666666666666,
523
+ "mean_length": 1680.8166666666666,
524
+ "std_reward": 95.54725125414242,
525
+ "std_length": 1741.913138780334,
526
+ "task_count": 4,
527
+ "macro_mean_reward": 65.86666666666666,
528
+ "macro_mean_length": 1680.8166666666666,
529
+ "distributed_eval": true
530
+ }
531
+ }
archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_1500.json ADDED
@@ -0,0 +1,532 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 5,
6
+ "mean_reward": -9.299999999999999,
7
+ "mean_length": 50.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ -9.299999999999999,
12
+ -9.299999999999999,
13
+ -9.299999999999999,
14
+ -9.299999999999999,
15
+ -9.299999999999999
16
+ ],
17
+ "episode_lengths": [
18
+ 50,
19
+ 50,
20
+ 50,
21
+ 50,
22
+ 50
23
+ ],
24
+ "decoded_action_hist": {
25
+ "1": 250
26
+ },
27
+ "fixed_episode_seeds": true,
28
+ "eval_seed": 42,
29
+ "episode_seeds": [
30
+ 42,
31
+ 43,
32
+ 44,
33
+ 45,
34
+ 46
35
+ ],
36
+ "episode_indices": [
37
+ 0,
38
+ 1,
39
+ 2,
40
+ 3,
41
+ 4
42
+ ]
43
+ },
44
+ "flappy/latency_1": {
45
+ "latency": 1,
46
+ "num_episodes": 5,
47
+ "mean_reward": -9.299999999999999,
48
+ "mean_length": 50.0,
49
+ "std_reward": 0.0,
50
+ "std_length": 0.0,
51
+ "episode_rewards": [
52
+ -9.299999999999999,
53
+ -9.299999999999999,
54
+ -9.299999999999999,
55
+ -9.299999999999999,
56
+ -9.299999999999999
57
+ ],
58
+ "episode_lengths": [
59
+ 50,
60
+ 50,
61
+ 50,
62
+ 50,
63
+ 50
64
+ ],
65
+ "decoded_action_hist": {
66
+ "0": 5,
67
+ "1": 245
68
+ },
69
+ "fixed_episode_seeds": true,
70
+ "eval_seed": 42,
71
+ "episode_seeds": [
72
+ 42,
73
+ 43,
74
+ 44,
75
+ 45,
76
+ 46
77
+ ],
78
+ "episode_indices": [
79
+ 0,
80
+ 1,
81
+ 2,
82
+ 3,
83
+ 4
84
+ ]
85
+ },
86
+ "flappy/latency_2": {
87
+ "latency": 2,
88
+ "num_episodes": 5,
89
+ "mean_reward": -9.299999999999999,
90
+ "mean_length": 50.0,
91
+ "std_reward": 0.0,
92
+ "std_length": 0.0,
93
+ "episode_rewards": [
94
+ -9.299999999999999,
95
+ -9.299999999999999,
96
+ -9.299999999999999,
97
+ -9.299999999999999,
98
+ -9.299999999999999
99
+ ],
100
+ "episode_lengths": [
101
+ 50,
102
+ 50,
103
+ 50,
104
+ 50,
105
+ 50
106
+ ],
107
+ "decoded_action_hist": {
108
+ "0": 10,
109
+ "1": 240
110
+ },
111
+ "fixed_episode_seeds": true,
112
+ "eval_seed": 42,
113
+ "episode_seeds": [
114
+ 42,
115
+ 43,
116
+ 44,
117
+ 45,
118
+ 46
119
+ ],
120
+ "episode_indices": [
121
+ 0,
122
+ 1,
123
+ 2,
124
+ 3,
125
+ 4
126
+ ]
127
+ },
128
+ "flappy/latency_3": {
129
+ "latency": 3,
130
+ "num_episodes": 5,
131
+ "mean_reward": -9.299999999999999,
132
+ "mean_length": 50.0,
133
+ "std_reward": 0.0,
134
+ "std_length": 0.0,
135
+ "episode_rewards": [
136
+ -9.299999999999999,
137
+ -9.299999999999999,
138
+ -9.299999999999999,
139
+ -9.299999999999999,
140
+ -9.299999999999999
141
+ ],
142
+ "episode_lengths": [
143
+ 50,
144
+ 50,
145
+ 50,
146
+ 50,
147
+ 50
148
+ ],
149
+ "decoded_action_hist": {
150
+ "0": 15,
151
+ "1": 235
152
+ },
153
+ "fixed_episode_seeds": true,
154
+ "eval_seed": 42,
155
+ "episode_seeds": [
156
+ 42,
157
+ 43,
158
+ 44,
159
+ 45,
160
+ 46
161
+ ],
162
+ "episode_indices": [
163
+ 0,
164
+ 1,
165
+ 2,
166
+ 3,
167
+ 4
168
+ ]
169
+ },
170
+ "flappy/latency_4": {
171
+ "latency": 4,
172
+ "num_episodes": 5,
173
+ "mean_reward": -8.7,
174
+ "mean_length": 50.0,
175
+ "std_reward": 0.0,
176
+ "std_length": 0.0,
177
+ "episode_rewards": [
178
+ -8.7,
179
+ -8.7,
180
+ -8.7,
181
+ -8.7,
182
+ -8.7
183
+ ],
184
+ "episode_lengths": [
185
+ 50,
186
+ 50,
187
+ 50,
188
+ 50,
189
+ 50
190
+ ],
191
+ "decoded_action_hist": {
192
+ "0": 20,
193
+ "1": 230
194
+ },
195
+ "fixed_episode_seeds": true,
196
+ "eval_seed": 42,
197
+ "episode_seeds": [
198
+ 42,
199
+ 43,
200
+ 44,
201
+ 45,
202
+ 46
203
+ ],
204
+ "episode_indices": [
205
+ 0,
206
+ 1,
207
+ 2,
208
+ 3,
209
+ 4
210
+ ]
211
+ },
212
+ "flappy/latency_5": {
213
+ "latency": 5,
214
+ "num_episodes": 5,
215
+ "mean_reward": -8.7,
216
+ "mean_length": 50.0,
217
+ "std_reward": 0.0,
218
+ "std_length": 0.0,
219
+ "episode_rewards": [
220
+ -8.7,
221
+ -8.7,
222
+ -8.7,
223
+ -8.7,
224
+ -8.7
225
+ ],
226
+ "episode_lengths": [
227
+ 50,
228
+ 50,
229
+ 50,
230
+ 50,
231
+ 50
232
+ ],
233
+ "decoded_action_hist": {
234
+ "0": 25,
235
+ "1": 225
236
+ },
237
+ "fixed_episode_seeds": true,
238
+ "eval_seed": 42,
239
+ "episode_seeds": [
240
+ 42,
241
+ 43,
242
+ 44,
243
+ 45,
244
+ 46
245
+ ],
246
+ "episode_indices": [
247
+ 0,
248
+ 1,
249
+ 2,
250
+ 3,
251
+ 4
252
+ ]
253
+ },
254
+ "demon_attack/latency_0": {
255
+ "latency": 0,
256
+ "num_episodes": 5,
257
+ "mean_reward": 155.0,
258
+ "mean_length": 3600.0,
259
+ "std_reward": 65.11528238439882,
260
+ "std_length": 0.0,
261
+ "episode_rewards": [
262
+ 150.0,
263
+ 60.0,
264
+ 150.0,
265
+ 265.0,
266
+ 150.0
267
+ ],
268
+ "episode_lengths": [
269
+ 3600,
270
+ 3600,
271
+ 3600,
272
+ 3600,
273
+ 3600
274
+ ],
275
+ "decoded_action_hist": {
276
+ "1": 17881,
277
+ "5": 119
278
+ },
279
+ "fixed_episode_seeds": true,
280
+ "eval_seed": 42,
281
+ "episode_seeds": [
282
+ 42,
283
+ 43,
284
+ 44,
285
+ 45,
286
+ 46
287
+ ],
288
+ "episode_indices": [
289
+ 0,
290
+ 1,
291
+ 2,
292
+ 3,
293
+ 4
294
+ ]
295
+ },
296
+ "demon_attack/latency_1": {
297
+ "latency": 1,
298
+ "num_episodes": 5,
299
+ "mean_reward": 464.0,
300
+ "mean_length": 3457.4,
301
+ "std_reward": 459.144857316294,
302
+ "std_length": 285.2,
303
+ "episode_rewards": [
304
+ 265.0,
305
+ 370.0,
306
+ 265.0,
307
+ 1360.0,
308
+ 60.0
309
+ ],
310
+ "episode_lengths": [
311
+ 3600,
312
+ 3600,
313
+ 3600,
314
+ 2887,
315
+ 3600
316
+ ],
317
+ "decoded_action_hist": {
318
+ "0": 5,
319
+ "1": 17047,
320
+ "5": 235
321
+ },
322
+ "fixed_episode_seeds": true,
323
+ "eval_seed": 42,
324
+ "episode_seeds": [
325
+ 42,
326
+ 43,
327
+ 44,
328
+ 45,
329
+ 46
330
+ ],
331
+ "episode_indices": [
332
+ 0,
333
+ 1,
334
+ 2,
335
+ 3,
336
+ 4
337
+ ]
338
+ },
339
+ "demon_attack/latency_2": {
340
+ "latency": 2,
341
+ "num_episodes": 5,
342
+ "mean_reward": 272.0,
343
+ "mean_length": 3415.0,
344
+ "std_reward": 231.9612036526798,
345
+ "std_length": 370.0,
346
+ "episode_rewards": [
347
+ 265.0,
348
+ 70.0,
349
+ 265.0,
350
+ 60.0,
351
+ 700.0
352
+ ],
353
+ "episode_lengths": [
354
+ 3600,
355
+ 3600,
356
+ 3600,
357
+ 3600,
358
+ 2675
359
+ ],
360
+ "decoded_action_hist": {
361
+ "0": 10,
362
+ "1": 16895,
363
+ "5": 167,
364
+ "3": 3
365
+ },
366
+ "fixed_episode_seeds": true,
367
+ "eval_seed": 42,
368
+ "episode_seeds": [
369
+ 42,
370
+ 43,
371
+ 44,
372
+ 45,
373
+ 46
374
+ ],
375
+ "episode_indices": [
376
+ 0,
377
+ 1,
378
+ 2,
379
+ 3,
380
+ 4
381
+ ]
382
+ },
383
+ "demon_attack/latency_3": {
384
+ "latency": 3,
385
+ "num_episodes": 5,
386
+ "mean_reward": 180.0,
387
+ "mean_length": 3600.0,
388
+ "std_reward": 75.2994023880668,
389
+ "std_length": 0.0,
390
+ "episode_rewards": [
391
+ 265.0,
392
+ 150.0,
393
+ 265.0,
394
+ 70.0,
395
+ 150.0
396
+ ],
397
+ "episode_lengths": [
398
+ 3600,
399
+ 3600,
400
+ 3600,
401
+ 3600,
402
+ 3600
403
+ ],
404
+ "decoded_action_hist": {
405
+ "0": 15,
406
+ "1": 17861,
407
+ "5": 124
408
+ },
409
+ "fixed_episode_seeds": true,
410
+ "eval_seed": 42,
411
+ "episode_seeds": [
412
+ 42,
413
+ 43,
414
+ 44,
415
+ 45,
416
+ 46
417
+ ],
418
+ "episode_indices": [
419
+ 0,
420
+ 1,
421
+ 2,
422
+ 3,
423
+ 4
424
+ ]
425
+ },
426
+ "demon_attack/latency_4": {
427
+ "latency": 4,
428
+ "num_episodes": 5,
429
+ "mean_reward": 158.0,
430
+ "mean_length": 3600.0,
431
+ "std_reward": 80.59776671843954,
432
+ "std_length": 0.0,
433
+ "episode_rewards": [
434
+ 250.0,
435
+ 70.0,
436
+ 250.0,
437
+ 70.0,
438
+ 150.0
439
+ ],
440
+ "episode_lengths": [
441
+ 3600,
442
+ 3600,
443
+ 3600,
444
+ 3600,
445
+ 3600
446
+ ],
447
+ "decoded_action_hist": {
448
+ "0": 20,
449
+ "1": 17578,
450
+ "5": 399,
451
+ "3": 3
452
+ },
453
+ "fixed_episode_seeds": true,
454
+ "eval_seed": 42,
455
+ "episode_seeds": [
456
+ 42,
457
+ 43,
458
+ 44,
459
+ 45,
460
+ 46
461
+ ],
462
+ "episode_indices": [
463
+ 0,
464
+ 1,
465
+ 2,
466
+ 3,
467
+ 4
468
+ ]
469
+ },
470
+ "demon_attack/latency_5": {
471
+ "latency": 5,
472
+ "num_episodes": 5,
473
+ "mean_reward": 157.0,
474
+ "mean_length": 3600.0,
475
+ "std_reward": 62.25752966509352,
476
+ "std_length": 0.0,
477
+ "episode_rewards": [
478
+ 150.0,
479
+ 150.0,
480
+ 150.0,
481
+ 265.0,
482
+ 70.0
483
+ ],
484
+ "episode_lengths": [
485
+ 3600,
486
+ 3600,
487
+ 3600,
488
+ 3600,
489
+ 3600
490
+ ],
491
+ "decoded_action_hist": {
492
+ "0": 25,
493
+ "1": 17824,
494
+ "5": 150,
495
+ "3": 1
496
+ },
497
+ "fixed_episode_seeds": true,
498
+ "eval_seed": 42,
499
+ "episode_seeds": [
500
+ 42,
501
+ 43,
502
+ 44,
503
+ 45,
504
+ 46
505
+ ],
506
+ "episode_indices": [
507
+ 0,
508
+ 1,
509
+ 2,
510
+ 3,
511
+ 4
512
+ ]
513
+ }
514
+ },
515
+ "aggregate": {
516
+ "stage": "mid_train",
517
+ "step": 1500,
518
+ "task": "cross_task",
519
+ "model_alias": "openvla",
520
+ "fixed_episode_seeds": true,
521
+ "eval_seed": 42,
522
+ "total_episodes": 60,
523
+ "mean_reward": 110.95,
524
+ "mean_length": 1797.7,
525
+ "std_reward": 210.75232501683104,
526
+ "std_length": 1753.7667870424884,
527
+ "task_count": 4,
528
+ "macro_mean_reward": 110.95,
529
+ "macro_mean_length": 1797.7,
530
+ "distributed_eval": true
531
+ }
532
+ }
archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_2000.json ADDED
@@ -0,0 +1,535 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 5,
6
+ "mean_reward": -9.299999999999999,
7
+ "mean_length": 50.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ -9.299999999999999,
12
+ -9.299999999999999,
13
+ -9.299999999999999,
14
+ -9.299999999999999,
15
+ -9.299999999999999
16
+ ],
17
+ "episode_lengths": [
18
+ 50,
19
+ 50,
20
+ 50,
21
+ 50,
22
+ 50
23
+ ],
24
+ "decoded_action_hist": {
25
+ "1": 250
26
+ },
27
+ "fixed_episode_seeds": true,
28
+ "eval_seed": 42,
29
+ "episode_seeds": [
30
+ 42,
31
+ 43,
32
+ 44,
33
+ 45,
34
+ 46
35
+ ],
36
+ "episode_indices": [
37
+ 0,
38
+ 1,
39
+ 2,
40
+ 3,
41
+ 4
42
+ ]
43
+ },
44
+ "flappy/latency_1": {
45
+ "latency": 1,
46
+ "num_episodes": 5,
47
+ "mean_reward": -9.299999999999999,
48
+ "mean_length": 50.0,
49
+ "std_reward": 0.0,
50
+ "std_length": 0.0,
51
+ "episode_rewards": [
52
+ -9.299999999999999,
53
+ -9.299999999999999,
54
+ -9.299999999999999,
55
+ -9.299999999999999,
56
+ -9.299999999999999
57
+ ],
58
+ "episode_lengths": [
59
+ 50,
60
+ 50,
61
+ 50,
62
+ 50,
63
+ 50
64
+ ],
65
+ "decoded_action_hist": {
66
+ "0": 5,
67
+ "1": 245
68
+ },
69
+ "fixed_episode_seeds": true,
70
+ "eval_seed": 42,
71
+ "episode_seeds": [
72
+ 42,
73
+ 43,
74
+ 44,
75
+ 45,
76
+ 46
77
+ ],
78
+ "episode_indices": [
79
+ 0,
80
+ 1,
81
+ 2,
82
+ 3,
83
+ 4
84
+ ]
85
+ },
86
+ "flappy/latency_2": {
87
+ "latency": 2,
88
+ "num_episodes": 5,
89
+ "mean_reward": -9.299999999999999,
90
+ "mean_length": 50.0,
91
+ "std_reward": 0.0,
92
+ "std_length": 0.0,
93
+ "episode_rewards": [
94
+ -9.299999999999999,
95
+ -9.299999999999999,
96
+ -9.299999999999999,
97
+ -9.299999999999999,
98
+ -9.299999999999999
99
+ ],
100
+ "episode_lengths": [
101
+ 50,
102
+ 50,
103
+ 50,
104
+ 50,
105
+ 50
106
+ ],
107
+ "decoded_action_hist": {
108
+ "0": 10,
109
+ "1": 240
110
+ },
111
+ "fixed_episode_seeds": true,
112
+ "eval_seed": 42,
113
+ "episode_seeds": [
114
+ 42,
115
+ 43,
116
+ 44,
117
+ 45,
118
+ 46
119
+ ],
120
+ "episode_indices": [
121
+ 0,
122
+ 1,
123
+ 2,
124
+ 3,
125
+ 4
126
+ ]
127
+ },
128
+ "flappy/latency_3": {
129
+ "latency": 3,
130
+ "num_episodes": 5,
131
+ "mean_reward": -9.299999999999999,
132
+ "mean_length": 50.0,
133
+ "std_reward": 0.0,
134
+ "std_length": 0.0,
135
+ "episode_rewards": [
136
+ -9.299999999999999,
137
+ -9.299999999999999,
138
+ -9.299999999999999,
139
+ -9.299999999999999,
140
+ -9.299999999999999
141
+ ],
142
+ "episode_lengths": [
143
+ 50,
144
+ 50,
145
+ 50,
146
+ 50,
147
+ 50
148
+ ],
149
+ "decoded_action_hist": {
150
+ "0": 15,
151
+ "1": 235
152
+ },
153
+ "fixed_episode_seeds": true,
154
+ "eval_seed": 42,
155
+ "episode_seeds": [
156
+ 42,
157
+ 43,
158
+ 44,
159
+ 45,
160
+ 46
161
+ ],
162
+ "episode_indices": [
163
+ 0,
164
+ 1,
165
+ 2,
166
+ 3,
167
+ 4
168
+ ]
169
+ },
170
+ "flappy/latency_4": {
171
+ "latency": 4,
172
+ "num_episodes": 5,
173
+ "mean_reward": -8.7,
174
+ "mean_length": 50.0,
175
+ "std_reward": 0.0,
176
+ "std_length": 0.0,
177
+ "episode_rewards": [
178
+ -8.7,
179
+ -8.7,
180
+ -8.7,
181
+ -8.7,
182
+ -8.7
183
+ ],
184
+ "episode_lengths": [
185
+ 50,
186
+ 50,
187
+ 50,
188
+ 50,
189
+ 50
190
+ ],
191
+ "decoded_action_hist": {
192
+ "0": 20,
193
+ "1": 230
194
+ },
195
+ "fixed_episode_seeds": true,
196
+ "eval_seed": 42,
197
+ "episode_seeds": [
198
+ 42,
199
+ 43,
200
+ 44,
201
+ 45,
202
+ 46
203
+ ],
204
+ "episode_indices": [
205
+ 0,
206
+ 1,
207
+ 2,
208
+ 3,
209
+ 4
210
+ ]
211
+ },
212
+ "flappy/latency_5": {
213
+ "latency": 5,
214
+ "num_episodes": 5,
215
+ "mean_reward": -8.7,
216
+ "mean_length": 50.0,
217
+ "std_reward": 0.0,
218
+ "std_length": 0.0,
219
+ "episode_rewards": [
220
+ -8.7,
221
+ -8.7,
222
+ -8.7,
223
+ -8.7,
224
+ -8.7
225
+ ],
226
+ "episode_lengths": [
227
+ 50,
228
+ 50,
229
+ 50,
230
+ 50,
231
+ 50
232
+ ],
233
+ "decoded_action_hist": {
234
+ "0": 25,
235
+ "1": 225
236
+ },
237
+ "fixed_episode_seeds": true,
238
+ "eval_seed": 42,
239
+ "episode_seeds": [
240
+ 42,
241
+ 43,
242
+ 44,
243
+ 45,
244
+ 46
245
+ ],
246
+ "episode_indices": [
247
+ 0,
248
+ 1,
249
+ 2,
250
+ 3,
251
+ 4
252
+ ]
253
+ },
254
+ "demon_attack/latency_0": {
255
+ "latency": 0,
256
+ "num_episodes": 5,
257
+ "mean_reward": 82.0,
258
+ "mean_length": 3600.0,
259
+ "std_reward": 29.257477676655586,
260
+ "std_length": 0.0,
261
+ "episode_rewards": [
262
+ 70.0,
263
+ 70.0,
264
+ 70.0,
265
+ 140.0,
266
+ 60.0
267
+ ],
268
+ "episode_lengths": [
269
+ 3600,
270
+ 3600,
271
+ 3600,
272
+ 3600,
273
+ 3600
274
+ ],
275
+ "decoded_action_hist": {
276
+ "1": 16258,
277
+ "5": 1738,
278
+ "3": 4
279
+ },
280
+ "fixed_episode_seeds": true,
281
+ "eval_seed": 42,
282
+ "episode_seeds": [
283
+ 42,
284
+ 43,
285
+ 44,
286
+ 45,
287
+ 46
288
+ ],
289
+ "episode_indices": [
290
+ 0,
291
+ 1,
292
+ 2,
293
+ 3,
294
+ 4
295
+ ]
296
+ },
297
+ "demon_attack/latency_1": {
298
+ "latency": 1,
299
+ "num_episodes": 5,
300
+ "mean_reward": 66.0,
301
+ "mean_length": 3600.0,
302
+ "std_reward": 4.898979485566356,
303
+ "std_length": 0.0,
304
+ "episode_rewards": [
305
+ 70.0,
306
+ 70.0,
307
+ 70.0,
308
+ 60.0,
309
+ 60.0
310
+ ],
311
+ "episode_lengths": [
312
+ 3600,
313
+ 3600,
314
+ 3600,
315
+ 3600,
316
+ 3600
317
+ ],
318
+ "decoded_action_hist": {
319
+ "0": 5,
320
+ "1": 14743,
321
+ "5": 3250,
322
+ "3": 2
323
+ },
324
+ "fixed_episode_seeds": true,
325
+ "eval_seed": 42,
326
+ "episode_seeds": [
327
+ 42,
328
+ 43,
329
+ 44,
330
+ 45,
331
+ 46
332
+ ],
333
+ "episode_indices": [
334
+ 0,
335
+ 1,
336
+ 2,
337
+ 3,
338
+ 4
339
+ ]
340
+ },
341
+ "demon_attack/latency_2": {
342
+ "latency": 2,
343
+ "num_episodes": 5,
344
+ "mean_reward": 64.0,
345
+ "mean_length": 3600.0,
346
+ "std_reward": 4.898979485566356,
347
+ "std_length": 0.0,
348
+ "episode_rewards": [
349
+ 70.0,
350
+ 60.0,
351
+ 70.0,
352
+ 60.0,
353
+ 60.0
354
+ ],
355
+ "episode_lengths": [
356
+ 3600,
357
+ 3600,
358
+ 3600,
359
+ 3600,
360
+ 3600
361
+ ],
362
+ "decoded_action_hist": {
363
+ "0": 10,
364
+ "1": 15543,
365
+ "5": 2441,
366
+ "3": 6
367
+ },
368
+ "fixed_episode_seeds": true,
369
+ "eval_seed": 42,
370
+ "episode_seeds": [
371
+ 42,
372
+ 43,
373
+ 44,
374
+ 45,
375
+ 46
376
+ ],
377
+ "episode_indices": [
378
+ 0,
379
+ 1,
380
+ 2,
381
+ 3,
382
+ 4
383
+ ]
384
+ },
385
+ "demon_attack/latency_3": {
386
+ "latency": 3,
387
+ "num_episodes": 5,
388
+ "mean_reward": 204.0,
389
+ "mean_length": 3600.0,
390
+ "std_reward": 139.08270920570968,
391
+ "std_length": 0.0,
392
+ "episode_rewards": [
393
+ 370.0,
394
+ 70.0,
395
+ 370.0,
396
+ 60.0,
397
+ 150.0
398
+ ],
399
+ "episode_lengths": [
400
+ 3600,
401
+ 3600,
402
+ 3600,
403
+ 3600,
404
+ 3600
405
+ ],
406
+ "decoded_action_hist": {
407
+ "0": 15,
408
+ "1": 16476,
409
+ "5": 1503,
410
+ "3": 6
411
+ },
412
+ "fixed_episode_seeds": true,
413
+ "eval_seed": 42,
414
+ "episode_seeds": [
415
+ 42,
416
+ 43,
417
+ 44,
418
+ 45,
419
+ 46
420
+ ],
421
+ "episode_indices": [
422
+ 0,
423
+ 1,
424
+ 2,
425
+ 3,
426
+ 4
427
+ ]
428
+ },
429
+ "demon_attack/latency_4": {
430
+ "latency": 4,
431
+ "num_episodes": 5,
432
+ "mean_reward": 96.0,
433
+ "mean_length": 3600.0,
434
+ "std_reward": 44.090815370097204,
435
+ "std_length": 0.0,
436
+ "episode_rewards": [
437
+ 150.0,
438
+ 60.0,
439
+ 150.0,
440
+ 60.0,
441
+ 60.0
442
+ ],
443
+ "episode_lengths": [
444
+ 3600,
445
+ 3600,
446
+ 3600,
447
+ 3600,
448
+ 3600
449
+ ],
450
+ "decoded_action_hist": {
451
+ "0": 20,
452
+ "1": 15177,
453
+ "5": 2801,
454
+ "3": 2
455
+ },
456
+ "fixed_episode_seeds": true,
457
+ "eval_seed": 42,
458
+ "episode_seeds": [
459
+ 42,
460
+ 43,
461
+ 44,
462
+ 45,
463
+ 46
464
+ ],
465
+ "episode_indices": [
466
+ 0,
467
+ 1,
468
+ 2,
469
+ 3,
470
+ 4
471
+ ]
472
+ },
473
+ "demon_attack/latency_5": {
474
+ "latency": 5,
475
+ "num_episodes": 5,
476
+ "mean_reward": 66.0,
477
+ "mean_length": 3600.0,
478
+ "std_reward": 4.898979485566356,
479
+ "std_length": 0.0,
480
+ "episode_rewards": [
481
+ 60.0,
482
+ 70.0,
483
+ 60.0,
484
+ 70.0,
485
+ 70.0
486
+ ],
487
+ "episode_lengths": [
488
+ 3600,
489
+ 3600,
490
+ 3600,
491
+ 3600,
492
+ 3600
493
+ ],
494
+ "decoded_action_hist": {
495
+ "0": 25,
496
+ "1": 15804,
497
+ "5": 2170,
498
+ "3": 1
499
+ },
500
+ "fixed_episode_seeds": true,
501
+ "eval_seed": 42,
502
+ "episode_seeds": [
503
+ 42,
504
+ 43,
505
+ 44,
506
+ 45,
507
+ 46
508
+ ],
509
+ "episode_indices": [
510
+ 0,
511
+ 1,
512
+ 2,
513
+ 3,
514
+ 4
515
+ ]
516
+ }
517
+ },
518
+ "aggregate": {
519
+ "stage": "mid_train",
520
+ "step": 2000,
521
+ "task": "cross_task",
522
+ "model_alias": "openvla",
523
+ "fixed_episode_seeds": true,
524
+ "eval_seed": 42,
525
+ "total_episodes": 60,
526
+ "mean_reward": 43.61666666666667,
527
+ "mean_length": 1825.0,
528
+ "std_reward": 76.51164217874879,
529
+ "std_length": 1775.0,
530
+ "task_count": 4,
531
+ "macro_mean_reward": 43.61666666666667,
532
+ "macro_mean_length": 1825.0,
533
+ "distributed_eval": true
534
+ }
535
+ }
archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/mid_train/step_500.json ADDED
@@ -0,0 +1,535 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 5,
6
+ "mean_reward": -9.299999999999999,
7
+ "mean_length": 50.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ -9.299999999999999,
12
+ -9.299999999999999,
13
+ -9.299999999999999,
14
+ -9.299999999999999,
15
+ -9.299999999999999
16
+ ],
17
+ "episode_lengths": [
18
+ 50,
19
+ 50,
20
+ 50,
21
+ 50,
22
+ 50
23
+ ],
24
+ "decoded_action_hist": {
25
+ "1": 250
26
+ },
27
+ "fixed_episode_seeds": true,
28
+ "eval_seed": 42,
29
+ "episode_seeds": [
30
+ 42,
31
+ 43,
32
+ 44,
33
+ 45,
34
+ 46
35
+ ],
36
+ "episode_indices": [
37
+ 0,
38
+ 1,
39
+ 2,
40
+ 3,
41
+ 4
42
+ ]
43
+ },
44
+ "flappy/latency_1": {
45
+ "latency": 1,
46
+ "num_episodes": 5,
47
+ "mean_reward": -9.299999999999999,
48
+ "mean_length": 50.0,
49
+ "std_reward": 0.0,
50
+ "std_length": 0.0,
51
+ "episode_rewards": [
52
+ -9.299999999999999,
53
+ -9.299999999999999,
54
+ -9.299999999999999,
55
+ -9.299999999999999,
56
+ -9.299999999999999
57
+ ],
58
+ "episode_lengths": [
59
+ 50,
60
+ 50,
61
+ 50,
62
+ 50,
63
+ 50
64
+ ],
65
+ "decoded_action_hist": {
66
+ "0": 5,
67
+ "1": 245
68
+ },
69
+ "fixed_episode_seeds": true,
70
+ "eval_seed": 42,
71
+ "episode_seeds": [
72
+ 42,
73
+ 43,
74
+ 44,
75
+ 45,
76
+ 46
77
+ ],
78
+ "episode_indices": [
79
+ 0,
80
+ 1,
81
+ 2,
82
+ 3,
83
+ 4
84
+ ]
85
+ },
86
+ "flappy/latency_2": {
87
+ "latency": 2,
88
+ "num_episodes": 5,
89
+ "mean_reward": -9.299999999999999,
90
+ "mean_length": 50.0,
91
+ "std_reward": 0.0,
92
+ "std_length": 0.0,
93
+ "episode_rewards": [
94
+ -9.299999999999999,
95
+ -9.299999999999999,
96
+ -9.299999999999999,
97
+ -9.299999999999999,
98
+ -9.299999999999999
99
+ ],
100
+ "episode_lengths": [
101
+ 50,
102
+ 50,
103
+ 50,
104
+ 50,
105
+ 50
106
+ ],
107
+ "decoded_action_hist": {
108
+ "0": 10,
109
+ "1": 240
110
+ },
111
+ "fixed_episode_seeds": true,
112
+ "eval_seed": 42,
113
+ "episode_seeds": [
114
+ 42,
115
+ 43,
116
+ 44,
117
+ 45,
118
+ 46
119
+ ],
120
+ "episode_indices": [
121
+ 0,
122
+ 1,
123
+ 2,
124
+ 3,
125
+ 4
126
+ ]
127
+ },
128
+ "flappy/latency_3": {
129
+ "latency": 3,
130
+ "num_episodes": 5,
131
+ "mean_reward": -9.299999999999999,
132
+ "mean_length": 50.0,
133
+ "std_reward": 0.0,
134
+ "std_length": 0.0,
135
+ "episode_rewards": [
136
+ -9.299999999999999,
137
+ -9.299999999999999,
138
+ -9.299999999999999,
139
+ -9.299999999999999,
140
+ -9.299999999999999
141
+ ],
142
+ "episode_lengths": [
143
+ 50,
144
+ 50,
145
+ 50,
146
+ 50,
147
+ 50
148
+ ],
149
+ "decoded_action_hist": {
150
+ "0": 15,
151
+ "1": 235
152
+ },
153
+ "fixed_episode_seeds": true,
154
+ "eval_seed": 42,
155
+ "episode_seeds": [
156
+ 42,
157
+ 43,
158
+ 44,
159
+ 45,
160
+ 46
161
+ ],
162
+ "episode_indices": [
163
+ 0,
164
+ 1,
165
+ 2,
166
+ 3,
167
+ 4
168
+ ]
169
+ },
170
+ "flappy/latency_4": {
171
+ "latency": 4,
172
+ "num_episodes": 5,
173
+ "mean_reward": -8.7,
174
+ "mean_length": 50.0,
175
+ "std_reward": 0.0,
176
+ "std_length": 0.0,
177
+ "episode_rewards": [
178
+ -8.7,
179
+ -8.7,
180
+ -8.7,
181
+ -8.7,
182
+ -8.7
183
+ ],
184
+ "episode_lengths": [
185
+ 50,
186
+ 50,
187
+ 50,
188
+ 50,
189
+ 50
190
+ ],
191
+ "decoded_action_hist": {
192
+ "0": 20,
193
+ "1": 230
194
+ },
195
+ "fixed_episode_seeds": true,
196
+ "eval_seed": 42,
197
+ "episode_seeds": [
198
+ 42,
199
+ 43,
200
+ 44,
201
+ 45,
202
+ 46
203
+ ],
204
+ "episode_indices": [
205
+ 0,
206
+ 1,
207
+ 2,
208
+ 3,
209
+ 4
210
+ ]
211
+ },
212
+ "flappy/latency_5": {
213
+ "latency": 5,
214
+ "num_episodes": 5,
215
+ "mean_reward": -8.7,
216
+ "mean_length": 50.0,
217
+ "std_reward": 0.0,
218
+ "std_length": 0.0,
219
+ "episode_rewards": [
220
+ -8.7,
221
+ -8.7,
222
+ -8.7,
223
+ -8.7,
224
+ -8.7
225
+ ],
226
+ "episode_lengths": [
227
+ 50,
228
+ 50,
229
+ 50,
230
+ 50,
231
+ 50
232
+ ],
233
+ "decoded_action_hist": {
234
+ "0": 25,
235
+ "1": 225
236
+ },
237
+ "fixed_episode_seeds": true,
238
+ "eval_seed": 42,
239
+ "episode_seeds": [
240
+ 42,
241
+ 43,
242
+ 44,
243
+ 45,
244
+ 46
245
+ ],
246
+ "episode_indices": [
247
+ 0,
248
+ 1,
249
+ 2,
250
+ 3,
251
+ 4
252
+ ]
253
+ },
254
+ "demon_attack/latency_0": {
255
+ "latency": 0,
256
+ "num_episodes": 5,
257
+ "mean_reward": 313.0,
258
+ "mean_length": 3600.0,
259
+ "std_reward": 58.787753826796276,
260
+ "std_length": 0.0,
261
+ "episode_rewards": [
262
+ 385.0,
263
+ 265.0,
264
+ 385.0,
265
+ 265.0,
266
+ 265.0
267
+ ],
268
+ "episode_lengths": [
269
+ 3600,
270
+ 3600,
271
+ 3600,
272
+ 3600,
273
+ 3600
274
+ ],
275
+ "decoded_action_hist": {
276
+ "1": 17778,
277
+ "5": 219,
278
+ "3": 3
279
+ },
280
+ "fixed_episode_seeds": true,
281
+ "eval_seed": 42,
282
+ "episode_seeds": [
283
+ 42,
284
+ 43,
285
+ 44,
286
+ 45,
287
+ 46
288
+ ],
289
+ "episode_indices": [
290
+ 0,
291
+ 1,
292
+ 2,
293
+ 3,
294
+ 4
295
+ ]
296
+ },
297
+ "demon_attack/latency_1": {
298
+ "latency": 1,
299
+ "num_episodes": 5,
300
+ "mean_reward": 230.0,
301
+ "mean_length": 2400.0,
302
+ "std_reward": 87.34987120768983,
303
+ "std_length": 984.3145838602616,
304
+ "episode_rewards": [
305
+ 265.0,
306
+ 250.0,
307
+ 265.0,
308
+ 310.0,
309
+ 60.0
310
+ ],
311
+ "episode_lengths": [
312
+ 1686,
313
+ 3600,
314
+ 1686,
315
+ 1428,
316
+ 3600
317
+ ],
318
+ "decoded_action_hist": {
319
+ "0": 5,
320
+ "1": 11865,
321
+ "5": 127,
322
+ "3": 3
323
+ },
324
+ "fixed_episode_seeds": true,
325
+ "eval_seed": 42,
326
+ "episode_seeds": [
327
+ 42,
328
+ 43,
329
+ 44,
330
+ 45,
331
+ 46
332
+ ],
333
+ "episode_indices": [
334
+ 0,
335
+ 1,
336
+ 2,
337
+ 3,
338
+ 4
339
+ ]
340
+ },
341
+ "demon_attack/latency_2": {
342
+ "latency": 2,
343
+ "num_episodes": 5,
344
+ "mean_reward": 249.0,
345
+ "mean_length": 2374.6,
346
+ "std_reward": 121.09500402576482,
347
+ "std_length": 1019.2883007275223,
348
+ "episode_rewards": [
349
+ 340.0,
350
+ 60.0,
351
+ 340.0,
352
+ 150.0,
353
+ 355.0
354
+ ],
355
+ "episode_lengths": [
356
+ 1380,
357
+ 3600,
358
+ 1380,
359
+ 3600,
360
+ 1913
361
+ ],
362
+ "decoded_action_hist": {
363
+ "0": 10,
364
+ "1": 11680,
365
+ "5": 171,
366
+ "3": 12
367
+ },
368
+ "fixed_episode_seeds": true,
369
+ "eval_seed": 42,
370
+ "episode_seeds": [
371
+ 42,
372
+ 43,
373
+ 44,
374
+ 45,
375
+ 46
376
+ ],
377
+ "episode_indices": [
378
+ 0,
379
+ 1,
380
+ 2,
381
+ 3,
382
+ 4
383
+ ]
384
+ },
385
+ "demon_attack/latency_3": {
386
+ "latency": 3,
387
+ "num_episodes": 5,
388
+ "mean_reward": 456.0,
389
+ "mean_length": 2611.6,
390
+ "std_reward": 362.91045727562056,
391
+ "std_length": 1009.3081987183102,
392
+ "episode_rewards": [
393
+ 900.0,
394
+ 140.0,
395
+ 900.0,
396
+ 190.0,
397
+ 150.0
398
+ ],
399
+ "episode_lengths": [
400
+ 2506,
401
+ 3600,
402
+ 2506,
403
+ 846,
404
+ 3600
405
+ ],
406
+ "decoded_action_hist": {
407
+ "0": 15,
408
+ "1": 12926,
409
+ "5": 112,
410
+ "3": 5
411
+ },
412
+ "fixed_episode_seeds": true,
413
+ "eval_seed": 42,
414
+ "episode_seeds": [
415
+ 42,
416
+ 43,
417
+ 44,
418
+ 45,
419
+ 46
420
+ ],
421
+ "episode_indices": [
422
+ 0,
423
+ 1,
424
+ 2,
425
+ 3,
426
+ 4
427
+ ]
428
+ },
429
+ "demon_attack/latency_4": {
430
+ "latency": 4,
431
+ "num_episodes": 5,
432
+ "mean_reward": 160.0,
433
+ "mean_length": 3600.0,
434
+ "std_reward": 111.71392035015153,
435
+ "std_length": 0.0,
436
+ "episode_rewards": [
437
+ 150.0,
438
+ 70.0,
439
+ 150.0,
440
+ 60.0,
441
+ 370.0
442
+ ],
443
+ "episode_lengths": [
444
+ 3600,
445
+ 3600,
446
+ 3600,
447
+ 3600,
448
+ 3600
449
+ ],
450
+ "decoded_action_hist": {
451
+ "0": 20,
452
+ "1": 17850,
453
+ "5": 126,
454
+ "3": 4
455
+ },
456
+ "fixed_episode_seeds": true,
457
+ "eval_seed": 42,
458
+ "episode_seeds": [
459
+ 42,
460
+ 43,
461
+ 44,
462
+ 45,
463
+ 46
464
+ ],
465
+ "episode_indices": [
466
+ 0,
467
+ 1,
468
+ 2,
469
+ 3,
470
+ 4
471
+ ]
472
+ },
473
+ "demon_attack/latency_5": {
474
+ "latency": 5,
475
+ "num_episodes": 5,
476
+ "mean_reward": 244.0,
477
+ "mean_length": 3315.0,
478
+ "std_reward": 144.72042012100434,
479
+ "std_length": 570.0,
480
+ "episode_rewards": [
481
+ 250.0,
482
+ 150.0,
483
+ 250.0,
484
+ 500.0,
485
+ 70.0
486
+ ],
487
+ "episode_lengths": [
488
+ 3600,
489
+ 3600,
490
+ 3600,
491
+ 2175,
492
+ 3600
493
+ ],
494
+ "decoded_action_hist": {
495
+ "0": 25,
496
+ "1": 16327,
497
+ "5": 222,
498
+ "3": 1
499
+ },
500
+ "fixed_episode_seeds": true,
501
+ "eval_seed": 42,
502
+ "episode_seeds": [
503
+ 42,
504
+ 43,
505
+ 44,
506
+ 45,
507
+ 46
508
+ ],
509
+ "episode_indices": [
510
+ 0,
511
+ 1,
512
+ 2,
513
+ 3,
514
+ 4
515
+ ]
516
+ }
517
+ },
518
+ "aggregate": {
519
+ "stage": "mid_train",
520
+ "step": 500,
521
+ "task": "cross_task",
522
+ "model_alias": "openvla",
523
+ "fixed_episode_seeds": true,
524
+ "eval_seed": 42,
525
+ "total_episodes": 60,
526
+ "mean_reward": 133.11666666666667,
527
+ "mean_length": 1516.7666666666667,
528
+ "std_reward": 200.9719666741829,
529
+ "std_length": 1604.3653508128657,
530
+ "task_count": 4,
531
+ "macro_mean_reward": 133.11666666666667,
532
+ "macro_mean_length": 1516.7666666666667,
533
+ "distributed_eval": true
534
+ }
535
+ }
archive/task-transfer/openvla_bridge_cross_defendtheline_zero_deadly_zero_exp3/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,1177 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 5,
6
+ "mean_reward": -9.299999999999999,
7
+ "mean_length": 50.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ -9.299999999999999,
12
+ -9.299999999999999,
13
+ -9.299999999999999,
14
+ -9.299999999999999,
15
+ -9.299999999999999
16
+ ],
17
+ "episode_lengths": [
18
+ 50,
19
+ 50,
20
+ 50,
21
+ 50,
22
+ 50
23
+ ],
24
+ "decoded_action_hist": {
25
+ "1": 250
26
+ },
27
+ "fixed_episode_seeds": true,
28
+ "eval_seed": 42,
29
+ "episode_seeds": [
30
+ 42,
31
+ 43,
32
+ 44,
33
+ 45,
34
+ 46
35
+ ],
36
+ "episode_indices": [
37
+ 0,
38
+ 1,
39
+ 2,
40
+ 3,
41
+ 4
42
+ ]
43
+ },
44
+ "flappy/latency_1": {
45
+ "latency": 1,
46
+ "num_episodes": 5,
47
+ "mean_reward": -9.299999999999999,
48
+ "mean_length": 50.0,
49
+ "std_reward": 0.0,
50
+ "std_length": 0.0,
51
+ "episode_rewards": [
52
+ -9.299999999999999,
53
+ -9.299999999999999,
54
+ -9.299999999999999,
55
+ -9.299999999999999,
56
+ -9.299999999999999
57
+ ],
58
+ "episode_lengths": [
59
+ 50,
60
+ 50,
61
+ 50,
62
+ 50,
63
+ 50
64
+ ],
65
+ "decoded_action_hist": {
66
+ "0": 5,
67
+ "1": 245
68
+ },
69
+ "fixed_episode_seeds": true,
70
+ "eval_seed": 42,
71
+ "episode_seeds": [
72
+ 42,
73
+ 43,
74
+ 44,
75
+ 45,
76
+ 46
77
+ ],
78
+ "episode_indices": [
79
+ 0,
80
+ 1,
81
+ 2,
82
+ 3,
83
+ 4
84
+ ]
85
+ },
86
+ "flappy/latency_2": {
87
+ "latency": 2,
88
+ "num_episodes": 5,
89
+ "mean_reward": -9.299999999999999,
90
+ "mean_length": 50.0,
91
+ "std_reward": 0.0,
92
+ "std_length": 0.0,
93
+ "episode_rewards": [
94
+ -9.299999999999999,
95
+ -9.299999999999999,
96
+ -9.299999999999999,
97
+ -9.299999999999999,
98
+ -9.299999999999999
99
+ ],
100
+ "episode_lengths": [
101
+ 50,
102
+ 50,
103
+ 50,
104
+ 50,
105
+ 50
106
+ ],
107
+ "decoded_action_hist": {
108
+ "0": 10,
109
+ "1": 240
110
+ },
111
+ "fixed_episode_seeds": true,
112
+ "eval_seed": 42,
113
+ "episode_seeds": [
114
+ 42,
115
+ 43,
116
+ 44,
117
+ 45,
118
+ 46
119
+ ],
120
+ "episode_indices": [
121
+ 0,
122
+ 1,
123
+ 2,
124
+ 3,
125
+ 4
126
+ ]
127
+ },
128
+ "flappy/latency_3": {
129
+ "latency": 3,
130
+ "num_episodes": 5,
131
+ "mean_reward": -9.299999999999999,
132
+ "mean_length": 50.0,
133
+ "std_reward": 0.0,
134
+ "std_length": 0.0,
135
+ "episode_rewards": [
136
+ -9.299999999999999,
137
+ -9.299999999999999,
138
+ -9.299999999999999,
139
+ -9.299999999999999,
140
+ -9.299999999999999
141
+ ],
142
+ "episode_lengths": [
143
+ 50,
144
+ 50,
145
+ 50,
146
+ 50,
147
+ 50
148
+ ],
149
+ "decoded_action_hist": {
150
+ "0": 15,
151
+ "1": 235
152
+ },
153
+ "fixed_episode_seeds": true,
154
+ "eval_seed": 42,
155
+ "episode_seeds": [
156
+ 42,
157
+ 43,
158
+ 44,
159
+ 45,
160
+ 46
161
+ ],
162
+ "episode_indices": [
163
+ 0,
164
+ 1,
165
+ 2,
166
+ 3,
167
+ 4
168
+ ]
169
+ },
170
+ "flappy/latency_4": {
171
+ "latency": 4,
172
+ "num_episodes": 5,
173
+ "mean_reward": -8.7,
174
+ "mean_length": 50.0,
175
+ "std_reward": 0.0,
176
+ "std_length": 0.0,
177
+ "episode_rewards": [
178
+ -8.7,
179
+ -8.7,
180
+ -8.7,
181
+ -8.7,
182
+ -8.7
183
+ ],
184
+ "episode_lengths": [
185
+ 50,
186
+ 50,
187
+ 50,
188
+ 50,
189
+ 50
190
+ ],
191
+ "decoded_action_hist": {
192
+ "0": 20,
193
+ "1": 230
194
+ },
195
+ "fixed_episode_seeds": true,
196
+ "eval_seed": 42,
197
+ "episode_seeds": [
198
+ 42,
199
+ 43,
200
+ 44,
201
+ 45,
202
+ 46
203
+ ],
204
+ "episode_indices": [
205
+ 0,
206
+ 1,
207
+ 2,
208
+ 3,
209
+ 4
210
+ ]
211
+ },
212
+ "flappy/latency_5": {
213
+ "latency": 5,
214
+ "num_episodes": 5,
215
+ "mean_reward": -8.7,
216
+ "mean_length": 50.0,
217
+ "std_reward": 0.0,
218
+ "std_length": 0.0,
219
+ "episode_rewards": [
220
+ -8.7,
221
+ -8.7,
222
+ -8.7,
223
+ -8.7,
224
+ -8.7
225
+ ],
226
+ "episode_lengths": [
227
+ 50,
228
+ 50,
229
+ 50,
230
+ 50,
231
+ 50
232
+ ],
233
+ "decoded_action_hist": {
234
+ "0": 25,
235
+ "1": 225
236
+ },
237
+ "fixed_episode_seeds": true,
238
+ "eval_seed": 42,
239
+ "episode_seeds": [
240
+ 42,
241
+ 43,
242
+ 44,
243
+ 45,
244
+ 46
245
+ ],
246
+ "episode_indices": [
247
+ 0,
248
+ 1,
249
+ 2,
250
+ 3,
251
+ 4
252
+ ]
253
+ },
254
+ "demon_attack/latency_0": {
255
+ "latency": 0,
256
+ "num_episodes": 5,
257
+ "mean_reward": 82.0,
258
+ "mean_length": 3600.0,
259
+ "std_reward": 29.257477676655586,
260
+ "std_length": 0.0,
261
+ "episode_rewards": [
262
+ 70.0,
263
+ 70.0,
264
+ 70.0,
265
+ 140.0,
266
+ 60.0
267
+ ],
268
+ "episode_lengths": [
269
+ 3600,
270
+ 3600,
271
+ 3600,
272
+ 3600,
273
+ 3600
274
+ ],
275
+ "decoded_action_hist": {
276
+ "1": 16258,
277
+ "5": 1738,
278
+ "3": 4
279
+ },
280
+ "fixed_episode_seeds": true,
281
+ "eval_seed": 42,
282
+ "episode_seeds": [
283
+ 42,
284
+ 43,
285
+ 44,
286
+ 45,
287
+ 46
288
+ ],
289
+ "episode_indices": [
290
+ 0,
291
+ 1,
292
+ 2,
293
+ 3,
294
+ 4
295
+ ]
296
+ },
297
+ "demon_attack/latency_1": {
298
+ "latency": 1,
299
+ "num_episodes": 5,
300
+ "mean_reward": 66.0,
301
+ "mean_length": 3600.0,
302
+ "std_reward": 4.898979485566356,
303
+ "std_length": 0.0,
304
+ "episode_rewards": [
305
+ 70.0,
306
+ 70.0,
307
+ 70.0,
308
+ 60.0,
309
+ 60.0
310
+ ],
311
+ "episode_lengths": [
312
+ 3600,
313
+ 3600,
314
+ 3600,
315
+ 3600,
316
+ 3600
317
+ ],
318
+ "decoded_action_hist": {
319
+ "0": 5,
320
+ "1": 14743,
321
+ "5": 3250,
322
+ "3": 2
323
+ },
324
+ "fixed_episode_seeds": true,
325
+ "eval_seed": 42,
326
+ "episode_seeds": [
327
+ 42,
328
+ 43,
329
+ 44,
330
+ 45,
331
+ 46
332
+ ],
333
+ "episode_indices": [
334
+ 0,
335
+ 1,
336
+ 2,
337
+ 3,
338
+ 4
339
+ ]
340
+ },
341
+ "demon_attack/latency_2": {
342
+ "latency": 2,
343
+ "num_episodes": 5,
344
+ "mean_reward": 64.0,
345
+ "mean_length": 3600.0,
346
+ "std_reward": 4.898979485566356,
347
+ "std_length": 0.0,
348
+ "episode_rewards": [
349
+ 70.0,
350
+ 60.0,
351
+ 70.0,
352
+ 60.0,
353
+ 60.0
354
+ ],
355
+ "episode_lengths": [
356
+ 3600,
357
+ 3600,
358
+ 3600,
359
+ 3600,
360
+ 3600
361
+ ],
362
+ "decoded_action_hist": {
363
+ "0": 10,
364
+ "1": 15543,
365
+ "5": 2441,
366
+ "3": 6
367
+ },
368
+ "fixed_episode_seeds": true,
369
+ "eval_seed": 42,
370
+ "episode_seeds": [
371
+ 42,
372
+ 43,
373
+ 44,
374
+ 45,
375
+ 46
376
+ ],
377
+ "episode_indices": [
378
+ 0,
379
+ 1,
380
+ 2,
381
+ 3,
382
+ 4
383
+ ]
384
+ },
385
+ "demon_attack/latency_3": {
386
+ "latency": 3,
387
+ "num_episodes": 5,
388
+ "mean_reward": 204.0,
389
+ "mean_length": 3600.0,
390
+ "std_reward": 139.08270920570968,
391
+ "std_length": 0.0,
392
+ "episode_rewards": [
393
+ 370.0,
394
+ 70.0,
395
+ 370.0,
396
+ 60.0,
397
+ 150.0
398
+ ],
399
+ "episode_lengths": [
400
+ 3600,
401
+ 3600,
402
+ 3600,
403
+ 3600,
404
+ 3600
405
+ ],
406
+ "decoded_action_hist": {
407
+ "0": 15,
408
+ "1": 16476,
409
+ "5": 1503,
410
+ "3": 6
411
+ },
412
+ "fixed_episode_seeds": true,
413
+ "eval_seed": 42,
414
+ "episode_seeds": [
415
+ 42,
416
+ 43,
417
+ 44,
418
+ 45,
419
+ 46
420
+ ],
421
+ "episode_indices": [
422
+ 0,
423
+ 1,
424
+ 2,
425
+ 3,
426
+ 4
427
+ ]
428
+ },
429
+ "demon_attack/latency_4": {
430
+ "latency": 4,
431
+ "num_episodes": 5,
432
+ "mean_reward": 96.0,
433
+ "mean_length": 3600.0,
434
+ "std_reward": 44.090815370097204,
435
+ "std_length": 0.0,
436
+ "episode_rewards": [
437
+ 150.0,
438
+ 60.0,
439
+ 150.0,
440
+ 60.0,
441
+ 60.0
442
+ ],
443
+ "episode_lengths": [
444
+ 3600,
445
+ 3600,
446
+ 3600,
447
+ 3600,
448
+ 3600
449
+ ],
450
+ "decoded_action_hist": {
451
+ "0": 20,
452
+ "1": 15177,
453
+ "5": 2801,
454
+ "3": 2
455
+ },
456
+ "fixed_episode_seeds": true,
457
+ "eval_seed": 42,
458
+ "episode_seeds": [
459
+ 42,
460
+ 43,
461
+ 44,
462
+ 45,
463
+ 46
464
+ ],
465
+ "episode_indices": [
466
+ 0,
467
+ 1,
468
+ 2,
469
+ 3,
470
+ 4
471
+ ]
472
+ },
473
+ "demon_attack/latency_5": {
474
+ "latency": 5,
475
+ "num_episodes": 5,
476
+ "mean_reward": 66.0,
477
+ "mean_length": 3600.0,
478
+ "std_reward": 4.898979485566356,
479
+ "std_length": 0.0,
480
+ "episode_rewards": [
481
+ 60.0,
482
+ 70.0,
483
+ 60.0,
484
+ 70.0,
485
+ 70.0
486
+ ],
487
+ "episode_lengths": [
488
+ 3600,
489
+ 3600,
490
+ 3600,
491
+ 3600,
492
+ 3600
493
+ ],
494
+ "decoded_action_hist": {
495
+ "0": 25,
496
+ "1": 15804,
497
+ "5": 2170,
498
+ "3": 1
499
+ },
500
+ "fixed_episode_seeds": true,
501
+ "eval_seed": 42,
502
+ "episode_seeds": [
503
+ 42,
504
+ 43,
505
+ 44,
506
+ 45,
507
+ 46
508
+ ],
509
+ "episode_indices": [
510
+ 0,
511
+ 1,
512
+ 2,
513
+ 3,
514
+ 4
515
+ ]
516
+ },
517
+ "defend_the_line/latency_0": {
518
+ "latency": 0,
519
+ "num_episodes": 20,
520
+ "mean_reward": 28.35,
521
+ "mean_length": 233.25,
522
+ "std_reward": 5.632716928800877,
523
+ "std_length": 51.23853530303145,
524
+ "episode_rewards": [
525
+ 26.0,
526
+ 29.0,
527
+ 26.0,
528
+ 32.0,
529
+ 21.0,
530
+ 29.0,
531
+ 27.0,
532
+ 31.0,
533
+ 21.0,
534
+ 31.0,
535
+ 40.0,
536
+ 34.0,
537
+ 33.0,
538
+ 19.0,
539
+ 33.0,
540
+ 23.0,
541
+ 21.0,
542
+ 38.0,
543
+ 25.0,
544
+ 28.0
545
+ ],
546
+ "episode_lengths": [
547
+ 209,
548
+ 219,
549
+ 208,
550
+ 284,
551
+ 174,
552
+ 225,
553
+ 219,
554
+ 245,
555
+ 162,
556
+ 255,
557
+ 348,
558
+ 307,
559
+ 274,
560
+ 139,
561
+ 257,
562
+ 188,
563
+ 199,
564
+ 307,
565
+ 208,
566
+ 238
567
+ ],
568
+ "decoded_action_hist": {
569
+ "[0, 1, 1]": 543,
570
+ "[0, 0, 1]": 3570,
571
+ "[1, 0, 1]": 552
572
+ },
573
+ "fixed_episode_seeds": true,
574
+ "eval_seed": 42,
575
+ "episode_seeds": [
576
+ 42,
577
+ 43,
578
+ 44,
579
+ 45,
580
+ 46,
581
+ 47,
582
+ 48,
583
+ 49,
584
+ 50,
585
+ 51,
586
+ 52,
587
+ 53,
588
+ 54,
589
+ 55,
590
+ 56,
591
+ 57,
592
+ 58,
593
+ 59,
594
+ 60,
595
+ 61
596
+ ],
597
+ "episode_indices": [
598
+ 0,
599
+ 1,
600
+ 2,
601
+ 3,
602
+ 4,
603
+ 5,
604
+ 6,
605
+ 7,
606
+ 8,
607
+ 9,
608
+ 10,
609
+ 11,
610
+ 12,
611
+ 13,
612
+ 14,
613
+ 15,
614
+ 16,
615
+ 17,
616
+ 18,
617
+ 19
618
+ ]
619
+ },
620
+ "defend_the_line/latency_2": {
621
+ "latency": 2,
622
+ "num_episodes": 20,
623
+ "mean_reward": 19.6,
624
+ "mean_length": 188.0,
625
+ "std_reward": 4.3977266854592045,
626
+ "std_length": 41.536730733171574,
627
+ "episode_rewards": [
628
+ 15.0,
629
+ 14.0,
630
+ 24.0,
631
+ 21.0,
632
+ 25.0,
633
+ 10.0,
634
+ 21.0,
635
+ 17.0,
636
+ 14.0,
637
+ 23.0,
638
+ 26.0,
639
+ 26.0,
640
+ 19.0,
641
+ 15.0,
642
+ 24.0,
643
+ 18.0,
644
+ 20.0,
645
+ 19.0,
646
+ 18.0,
647
+ 23.0
648
+ ],
649
+ "episode_lengths": [
650
+ 136,
651
+ 143,
652
+ 206,
653
+ 199,
654
+ 213,
655
+ 109,
656
+ 233,
657
+ 177,
658
+ 125,
659
+ 228,
660
+ 254,
661
+ 248,
662
+ 170,
663
+ 133,
664
+ 221,
665
+ 194,
666
+ 190,
667
+ 199,
668
+ 158,
669
+ 224
670
+ ],
671
+ "decoded_action_hist": {
672
+ "[0, 0, 0]": 40,
673
+ "[0, 1, 1]": 831,
674
+ "[0, 0, 1]": 2098,
675
+ "[1, 0, 1]": 791
676
+ },
677
+ "fixed_episode_seeds": true,
678
+ "eval_seed": 42,
679
+ "episode_seeds": [
680
+ 42,
681
+ 43,
682
+ 44,
683
+ 45,
684
+ 46,
685
+ 47,
686
+ 48,
687
+ 49,
688
+ 50,
689
+ 51,
690
+ 52,
691
+ 53,
692
+ 54,
693
+ 55,
694
+ 56,
695
+ 57,
696
+ 58,
697
+ 59,
698
+ 60,
699
+ 61
700
+ ],
701
+ "episode_indices": [
702
+ 0,
703
+ 1,
704
+ 2,
705
+ 3,
706
+ 4,
707
+ 5,
708
+ 6,
709
+ 7,
710
+ 8,
711
+ 9,
712
+ 10,
713
+ 11,
714
+ 12,
715
+ 13,
716
+ 14,
717
+ 15,
718
+ 16,
719
+ 17,
720
+ 18,
721
+ 19
722
+ ]
723
+ },
724
+ "defend_the_line/latency_4": {
725
+ "latency": 4,
726
+ "num_episodes": 20,
727
+ "mean_reward": 16.25,
728
+ "mean_length": 179.55,
729
+ "std_reward": 4.794528131109463,
730
+ "std_length": 53.87900797156532,
731
+ "episode_rewards": [
732
+ 12.0,
733
+ 11.0,
734
+ 17.0,
735
+ 21.0,
736
+ 16.0,
737
+ 13.0,
738
+ 18.0,
739
+ 13.0,
740
+ 13.0,
741
+ 12.0,
742
+ 22.0,
743
+ 14.0,
744
+ 17.0,
745
+ 12.0,
746
+ 19.0,
747
+ 10.0,
748
+ 16.0,
749
+ 31.0,
750
+ 18.0,
751
+ 20.0
752
+ ],
753
+ "episode_lengths": [
754
+ 140,
755
+ 125,
756
+ 159,
757
+ 202,
758
+ 197,
759
+ 155,
760
+ 182,
761
+ 147,
762
+ 124,
763
+ 140,
764
+ 251,
765
+ 161,
766
+ 182,
767
+ 147,
768
+ 205,
769
+ 121,
770
+ 184,
771
+ 362,
772
+ 182,
773
+ 225
774
+ ],
775
+ "decoded_action_hist": {
776
+ "[0, 0, 0]": 80,
777
+ "[0, 1, 1]": 988,
778
+ "[0, 0, 1]": 1609,
779
+ "[1, 0, 1]": 914
780
+ },
781
+ "fixed_episode_seeds": true,
782
+ "eval_seed": 42,
783
+ "episode_seeds": [
784
+ 42,
785
+ 43,
786
+ 44,
787
+ 45,
788
+ 46,
789
+ 47,
790
+ 48,
791
+ 49,
792
+ 50,
793
+ 51,
794
+ 52,
795
+ 53,
796
+ 54,
797
+ 55,
798
+ 56,
799
+ 57,
800
+ 58,
801
+ 59,
802
+ 60,
803
+ 61
804
+ ],
805
+ "episode_indices": [
806
+ 0,
807
+ 1,
808
+ 2,
809
+ 3,
810
+ 4,
811
+ 5,
812
+ 6,
813
+ 7,
814
+ 8,
815
+ 9,
816
+ 10,
817
+ 11,
818
+ 12,
819
+ 13,
820
+ 14,
821
+ 15,
822
+ 16,
823
+ 17,
824
+ 18,
825
+ 19
826
+ ]
827
+ },
828
+ "deadly_corridor/latency_0": {
829
+ "latency": 0,
830
+ "num_episodes": 20,
831
+ "mean_reward": 540.8966720581054,
832
+ "mean_length": 24.25,
833
+ "std_reward": 407.3320167056893,
834
+ "std_length": 5.337368265353254,
835
+ "episode_rewards": [
836
+ 417.030517578125,
837
+ 369.29583740234375,
838
+ 439.2748107910156,
839
+ 464.7837829589844,
840
+ 481.87049865722656,
841
+ 411.6549835205078,
842
+ 2286.3500213623047,
843
+ 471.70501708984375,
844
+ 431.1736755371094,
845
+ 368.65252685546875,
846
+ 412.97520446777344,
847
+ 439.42803955078125,
848
+ 470.2660675048828,
849
+ 507.5508575439453,
850
+ 454.8498229980469,
851
+ 525.7742309570312,
852
+ 321.8816680908203,
853
+ 351.98521423339844,
854
+ 679.7676544189453,
855
+ 511.6630096435547
856
+ ],
857
+ "episode_lengths": [
858
+ 22,
859
+ 21,
860
+ 23,
861
+ 23,
862
+ 24,
863
+ 22,
864
+ 45,
865
+ 23,
866
+ 22,
867
+ 20,
868
+ 25,
869
+ 23,
870
+ 24,
871
+ 27,
872
+ 24,
873
+ 25,
874
+ 18,
875
+ 20,
876
+ 29,
877
+ 25
878
+ ],
879
+ "decoded_action_hist": {
880
+ "[0, 1, 0, 0, 0, 1, 0]": 40,
881
+ "[0, 1, 1, 1, 0, 1, 0]": 87,
882
+ "[0, 1, 1, 1, 0, 0, 0]": 223,
883
+ "[0, 0, 1, 1, 0, 0, 0]": 15,
884
+ "[0, 1, 0, 1, 0, 0, 0]": 88,
885
+ "[0, 1, 0, 1, 0, 0, 1]": 14,
886
+ "[0, 1, 1, 0, 0, 0, 0]": 1,
887
+ "[0, 0, 0, 1, 0, 1, 0]": 1,
888
+ "[0, 1, 1, 1, 0, 0, 1]": 2,
889
+ "[0, 0, 0, 1, 0, 0, 0]": 6,
890
+ "[1, 0, 0, 1, 0, 0, 0]": 3,
891
+ "[1, 0, 1, 1, 0, 0, 0]": 2,
892
+ "[1, 0, 0, 1, 0, 1, 0]": 1,
893
+ "[0, 0, 1, 1, 0, 1, 0]": 1,
894
+ "[0, 1, 0, 1, 0, 1, 0]": 1
895
+ },
896
+ "fixed_episode_seeds": true,
897
+ "eval_seed": 42,
898
+ "episode_seeds": [
899
+ 42,
900
+ 43,
901
+ 44,
902
+ 45,
903
+ 46,
904
+ 47,
905
+ 48,
906
+ 49,
907
+ 50,
908
+ 51,
909
+ 52,
910
+ 53,
911
+ 54,
912
+ 55,
913
+ 56,
914
+ 57,
915
+ 58,
916
+ 59,
917
+ 60,
918
+ 61
919
+ ],
920
+ "episode_indices": [
921
+ 0,
922
+ 1,
923
+ 2,
924
+ 3,
925
+ 4,
926
+ 5,
927
+ 6,
928
+ 7,
929
+ 8,
930
+ 9,
931
+ 10,
932
+ 11,
933
+ 12,
934
+ 13,
935
+ 14,
936
+ 15,
937
+ 16,
938
+ 17,
939
+ 18,
940
+ 19
941
+ ]
942
+ },
943
+ "deadly_corridor/latency_2": {
944
+ "latency": 2,
945
+ "num_episodes": 20,
946
+ "mean_reward": 262.87070693969724,
947
+ "mean_length": 19.75,
948
+ "std_reward": 124.2434788467642,
949
+ "std_length": 5.3091901453988255,
950
+ "episode_rewards": [
951
+ 106.18881225585938,
952
+ 51.31547546386719,
953
+ 267.7543029785156,
954
+ 294.97987365722656,
955
+ 325.9807891845703,
956
+ 209.5471954345703,
957
+ 88.69207763671875,
958
+ 451.2413024902344,
959
+ 86.4166259765625,
960
+ 253.77011108398438,
961
+ 320.5774230957031,
962
+ 324.5970458984375,
963
+ 519.2172088623047,
964
+ 339.3174133300781,
965
+ 360.599853515625,
966
+ 307.4403076171875,
967
+ 261.3071594238281,
968
+ 310.2710876464844,
969
+ 313.8670349121094,
970
+ 64.33303833007812
971
+ ],
972
+ "episode_lengths": [
973
+ 13,
974
+ 12,
975
+ 19,
976
+ 22,
977
+ 21,
978
+ 17,
979
+ 12,
980
+ 25,
981
+ 12,
982
+ 19,
983
+ 22,
984
+ 25,
985
+ 29,
986
+ 25,
987
+ 21,
988
+ 20,
989
+ 19,
990
+ 24,
991
+ 27,
992
+ 11
993
+ ],
994
+ "decoded_action_hist": {
995
+ "[0, 0, 0, 0, 0, 0, 0]": 40,
996
+ "[0, 1, 0, 0, 0, 1, 0]": 40,
997
+ "[0, 1, 1, 1, 0, 1, 0]": 122,
998
+ "[0, 1, 1, 1, 0, 0, 0]": 187,
999
+ "[0, 1, 0, 1, 0, 0, 0]": 2,
1000
+ "[0, 1, 1, 0, 1, 0, 0]": 1,
1001
+ "[0, 1, 1, 0, 0, 0, 0]": 3
1002
+ },
1003
+ "fixed_episode_seeds": true,
1004
+ "eval_seed": 42,
1005
+ "episode_seeds": [
1006
+ 42,
1007
+ 43,
1008
+ 44,
1009
+ 45,
1010
+ 46,
1011
+ 47,
1012
+ 48,
1013
+ 49,
1014
+ 50,
1015
+ 51,
1016
+ 52,
1017
+ 53,
1018
+ 54,
1019
+ 55,
1020
+ 56,
1021
+ 57,
1022
+ 58,
1023
+ 59,
1024
+ 60,
1025
+ 61
1026
+ ],
1027
+ "episode_indices": [
1028
+ 0,
1029
+ 1,
1030
+ 2,
1031
+ 3,
1032
+ 4,
1033
+ 5,
1034
+ 6,
1035
+ 7,
1036
+ 8,
1037
+ 9,
1038
+ 10,
1039
+ 11,
1040
+ 12,
1041
+ 13,
1042
+ 14,
1043
+ 15,
1044
+ 16,
1045
+ 17,
1046
+ 18,
1047
+ 19
1048
+ ]
1049
+ },
1050
+ "deadly_corridor/latency_4": {
1051
+ "latency": 4,
1052
+ "num_episodes": 20,
1053
+ "mean_reward": 118.65168380737305,
1054
+ "mean_length": 20.05,
1055
+ "std_reward": 107.15731153590515,
1056
+ "std_length": 8.399255919425244,
1057
+ "episode_rewards": [
1058
+ 128.03665161132812,
1059
+ 23.916595458984375,
1060
+ 176.50045776367188,
1061
+ 26.05596923828125,
1062
+ 394.8277282714844,
1063
+ 138.68695068359375,
1064
+ 17.6407470703125,
1065
+ 346.8414611816406,
1066
+ 27.210739135742188,
1067
+ 50.12123107910156,
1068
+ 160.34852600097656,
1069
+ 53.54322814941406,
1070
+ 228.23159790039062,
1071
+ 62.88116455078125,
1072
+ 121.09461975097656,
1073
+ 225.57135009765625,
1074
+ 98.22915649414062,
1075
+ 47.8914794921875,
1076
+ 34.82347106933594,
1077
+ 10.580551147460938
1078
+ ],
1079
+ "episode_lengths": [
1080
+ 18,
1081
+ 12,
1082
+ 26,
1083
+ 13,
1084
+ 25,
1085
+ 25,
1086
+ 12,
1087
+ 30,
1088
+ 12,
1089
+ 13,
1090
+ 28,
1091
+ 13,
1092
+ 32,
1093
+ 18,
1094
+ 27,
1095
+ 41,
1096
+ 19,
1097
+ 13,
1098
+ 13,
1099
+ 11
1100
+ ],
1101
+ "decoded_action_hist": {
1102
+ "[0, 0, 0, 0, 0, 0, 0]": 80,
1103
+ "[0, 1, 0, 0, 0, 1, 0]": 40,
1104
+ "[0, 1, 1, 1, 0, 1, 0]": 137,
1105
+ "[0, 1, 1, 1, 0, 0, 0]": 136,
1106
+ "[0, 0, 1, 1, 0, 0, 0]": 2,
1107
+ "[0, 1, 1, 0, 1, 0, 0]": 1,
1108
+ "[0, 1, 1, 0, 0, 0, 0]": 2,
1109
+ "[0, 1, 0, 1, 0, 0, 0]": 2,
1110
+ "[0, 1, 1, 1, 0, 0, 1]": 1
1111
+ },
1112
+ "fixed_episode_seeds": true,
1113
+ "eval_seed": 42,
1114
+ "episode_seeds": [
1115
+ 42,
1116
+ 43,
1117
+ 44,
1118
+ 45,
1119
+ 46,
1120
+ 47,
1121
+ 48,
1122
+ 49,
1123
+ 50,
1124
+ 51,
1125
+ 52,
1126
+ 53,
1127
+ 54,
1128
+ 55,
1129
+ 56,
1130
+ 57,
1131
+ 58,
1132
+ 59,
1133
+ 60,
1134
+ 61
1135
+ ],
1136
+ "episode_indices": [
1137
+ 0,
1138
+ 1,
1139
+ 2,
1140
+ 3,
1141
+ 4,
1142
+ 5,
1143
+ 6,
1144
+ 7,
1145
+ 8,
1146
+ 9,
1147
+ 10,
1148
+ 11,
1149
+ 12,
1150
+ 13,
1151
+ 14,
1152
+ 15,
1153
+ 16,
1154
+ 17,
1155
+ 18,
1156
+ 19
1157
+ ]
1158
+ }
1159
+ },
1160
+ "aggregate": {
1161
+ "stage": "post_train",
1162
+ "step": 2000,
1163
+ "task": "cross_task",
1164
+ "model_alias": "openvla",
1165
+ "fixed_episode_seeds": true,
1166
+ "eval_seed": 42,
1167
+ "total_episodes": 180,
1168
+ "mean_reward": 124.16322920057509,
1169
+ "mean_length": 682.2055555555555,
1170
+ "std_reward": 224.75510410624398,
1171
+ "std_length": 1307.493742059476,
1172
+ "task_count": 4,
1173
+ "macro_mean_reward": 83.88994793362086,
1174
+ "macro_mean_length": 1253.6027777777776,
1175
+ "distributed_eval": true
1176
+ }
1177
+ }
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/.gitattributes ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ wandb/wandb/run-20260923_065340-67ykhdhp/run-67ykhdhp.wandb filter=lfs diff=lfs merge=lfs -text
37
+ wandb/wandb/run-20260923_105406-7ujo2tgu/run-7ujo2tgu.wandb filter=lfs diff=lfs merge=lfs -text
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/_initialization/StarVLA_Qwen3VL-OFT-Bridge-RT-1/checkpoints/steps_5000_pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:371cb744227687bb99bcad7f9ff2250cf06da75631359ad3eba4c6bc52570607
3
+ size 9785060316
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/steps_15000_state/optimizer.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e8b2dc7f1a406b093db1b4a03d3ae0dc2d77415802548498921dd5b5668fe8ed
3
+ size 18276885098
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/steps_15000_state/random_states_0.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f041f3f444886eb524567d2ff1df968b6f86a749c7ad6ad18b8d1478843cfb4
3
+ size 14821
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/eval/mid_train/step_1000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "defend_the_line/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 5,
6
+ "mean_reward": 8.8,
7
+ "mean_length": 122.4,
8
+ "std_reward": 2.4,
9
+ "std_length": 33.27221062688802,
10
+ "episode_rewards": [
11
+ 7.0,
12
+ 7.0,
13
+ 7.0,
14
+ 10.0,
15
+ 13.0
16
+ ],
17
+ "episode_lengths": [
18
+ 91,
19
+ 93,
20
+ 107,
21
+ 144,
22
+ 177
23
+ ],
24
+ "decoded_action_hist": {
25
+ "[0, 0, 1]": 611,
26
+ "[0, 1, 1]": 1
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1000,
49
+ "task": "cross_task",
50
+ "model_alias": "openvla",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 8.8,
55
+ "mean_length": 122.4,
56
+ "std_reward": 2.4,
57
+ "std_length": 33.27221062688802,
58
+ "task_count": 6,
59
+ "macro_mean_reward": 8.8,
60
+ "macro_mean_length": 122.4,
61
+ "distributed_eval": false
62
+ }
63
+ }
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/eval/post_train/step_15000.json ADDED
@@ -0,0 +1,993 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 436.44000000002427,
7
+ "mean_length": 3534.25,
8
+ "std_reward": 35.56861537929826,
9
+ "std_length": 286.59760553779927,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566,
16
+ 444.60000000002566,
17
+ 444.60000000002566,
18
+ 444.60000000002566,
19
+ 444.60000000002566,
20
+ 444.60000000002566,
21
+ 444.60000000002566,
22
+ 444.60000000002566,
23
+ 444.60000000002566,
24
+ 444.60000000002566,
25
+ 444.60000000002566,
26
+ 281.3999999999965,
27
+ 444.60000000002566,
28
+ 444.60000000002566,
29
+ 444.60000000002566,
30
+ 444.60000000002566
31
+ ],
32
+ "episode_lengths": [
33
+ 3600,
34
+ 3600,
35
+ 3600,
36
+ 3600,
37
+ 3600,
38
+ 3600,
39
+ 3600,
40
+ 3600,
41
+ 3600,
42
+ 3600,
43
+ 3600,
44
+ 3600,
45
+ 3600,
46
+ 3600,
47
+ 3600,
48
+ 2285,
49
+ 3600,
50
+ 3600,
51
+ 3600,
52
+ 3600
53
+ ],
54
+ "decoded_action_hist": {
55
+ "0": 65076,
56
+ "1": 5609
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ },
105
+ "flappy/latency_2": {
106
+ "latency": 2,
107
+ "num_episodes": 20,
108
+ "mean_reward": 396.61500000001854,
109
+ "mean_length": 3214.8,
110
+ "std_reward": 86.11668987485571,
111
+ "std_length": 692.3197671596558,
112
+ "episode_rewards": [
113
+ 444.60000000002566,
114
+ 444.60000000002566,
115
+ 444.60000000002566,
116
+ 199.89999999999435,
117
+ 444.60000000002566,
118
+ 444.60000000002566,
119
+ 185.89999999999498,
120
+ 444.60000000002566,
121
+ 444.60000000002566,
122
+ 444.60000000002566,
123
+ 444.60000000002566,
124
+ 444.60000000002566,
125
+ 311.9000000000021,
126
+ 444.60000000002566,
127
+ 444.60000000002566,
128
+ 444.60000000002566,
129
+ 428.4000000000229,
130
+ 444.60000000002566,
131
+ 255.89999999999208,
132
+ 325.9000000000046
133
+ ],
134
+ "episode_lengths": [
135
+ 3600,
136
+ 3600,
137
+ 3600,
138
+ 1632,
139
+ 3600,
140
+ 3600,
141
+ 1519,
142
+ 3600,
143
+ 3600,
144
+ 3600,
145
+ 3600,
146
+ 3600,
147
+ 2536,
148
+ 3600,
149
+ 3600,
150
+ 3600,
151
+ 3476,
152
+ 3600,
153
+ 2084,
154
+ 2649
155
+ ],
156
+ "decoded_action_hist": {
157
+ "0": 51213,
158
+ "1": 13083
159
+ },
160
+ "fixed_episode_seeds": true,
161
+ "eval_seed": 42,
162
+ "episode_seeds": [
163
+ 42,
164
+ 43,
165
+ 44,
166
+ 45,
167
+ 46,
168
+ 47,
169
+ 48,
170
+ 49,
171
+ 50,
172
+ 51,
173
+ 52,
174
+ 53,
175
+ 54,
176
+ 55,
177
+ 56,
178
+ 57,
179
+ 58,
180
+ 59,
181
+ 60,
182
+ 61
183
+ ],
184
+ "episode_indices": [
185
+ 0,
186
+ 1,
187
+ 2,
188
+ 3,
189
+ 4,
190
+ 5,
191
+ 6,
192
+ 7,
193
+ 8,
194
+ 9,
195
+ 10,
196
+ 11,
197
+ 12,
198
+ 13,
199
+ 14,
200
+ 15,
201
+ 16,
202
+ 17,
203
+ 18,
204
+ 19
205
+ ]
206
+ },
207
+ "flappy/latency_4": {
208
+ "latency": 4,
209
+ "num_episodes": 20,
210
+ "mean_reward": 325.9350000000128,
211
+ "mean_length": 2645.2,
212
+ "std_reward": 142.96879825683538,
213
+ "std_length": 1151.4947068918727,
214
+ "episode_rewards": [
215
+ 165.09999999999593,
216
+ 444.60000000002566,
217
+ 431.1000000000233,
218
+ 348.9000000000087,
219
+ 444.60000000002566,
220
+ 20.800000000000036,
221
+ 444.60000000002566,
222
+ 444.60000000002566,
223
+ 444.60000000002566,
224
+ 444.60000000002566,
225
+ 223.89999999999327,
226
+ 265.8999999999939,
227
+ 444.60000000002566,
228
+ 444.60000000002566,
229
+ 151.09999999999656,
230
+ 134.59999999999727,
231
+ 214.89999999999367,
232
+ 116.39999999999809,
233
+ 444.60000000002566,
234
+ 444.60000000002566
235
+ ],
236
+ "episode_lengths": [
237
+ 1347,
238
+ 3600,
239
+ 3494,
240
+ 2834,
241
+ 3600,
242
+ 183,
243
+ 3600,
244
+ 3600,
245
+ 3600,
246
+ 3600,
247
+ 1827,
248
+ 2166,
249
+ 3600,
250
+ 3600,
251
+ 1234,
252
+ 1105,
253
+ 1755,
254
+ 959,
255
+ 3600,
256
+ 3600
257
+ ],
258
+ "decoded_action_hist": {
259
+ "0": 43356,
260
+ "1": 9548
261
+ },
262
+ "fixed_episode_seeds": true,
263
+ "eval_seed": 42,
264
+ "episode_seeds": [
265
+ 42,
266
+ 43,
267
+ 44,
268
+ 45,
269
+ 46,
270
+ 47,
271
+ 48,
272
+ 49,
273
+ 50,
274
+ 51,
275
+ 52,
276
+ 53,
277
+ 54,
278
+ 55,
279
+ 56,
280
+ 57,
281
+ 58,
282
+ 59,
283
+ 60,
284
+ 61
285
+ ],
286
+ "episode_indices": [
287
+ 0,
288
+ 1,
289
+ 2,
290
+ 3,
291
+ 4,
292
+ 5,
293
+ 6,
294
+ 7,
295
+ 8,
296
+ 9,
297
+ 10,
298
+ 11,
299
+ 12,
300
+ 13,
301
+ 14,
302
+ 15,
303
+ 16,
304
+ 17,
305
+ 18,
306
+ 19
307
+ ]
308
+ },
309
+ "demon_attack/latency_0": {
310
+ "latency": 0,
311
+ "num_episodes": 20,
312
+ "mean_reward": 7097.0,
313
+ "mean_length": 3600.0,
314
+ "std_reward": 338.8672306376053,
315
+ "std_length": 0.0,
316
+ "episode_rewards": [
317
+ 7420.0,
318
+ 7090.0,
319
+ 7420.0,
320
+ 7090.0,
321
+ 7570.0,
322
+ 7000.0,
323
+ 6440.0,
324
+ 7000.0,
325
+ 7180.0,
326
+ 7000.0,
327
+ 7420.0,
328
+ 7000.0,
329
+ 6460.0,
330
+ 7000.0,
331
+ 7420.0,
332
+ 7810.0,
333
+ 7000.0,
334
+ 7000.0,
335
+ 7000.0,
336
+ 6620.0
337
+ ],
338
+ "episode_lengths": [
339
+ 3600,
340
+ 3600,
341
+ 3600,
342
+ 3600,
343
+ 3600,
344
+ 3600,
345
+ 3600,
346
+ 3600,
347
+ 3600,
348
+ 3600,
349
+ 3600,
350
+ 3600,
351
+ 3600,
352
+ 3600,
353
+ 3600,
354
+ 3600,
355
+ 3600,
356
+ 3600,
357
+ 3600,
358
+ 3600
359
+ ],
360
+ "decoded_action_hist": {
361
+ "3": 12448,
362
+ "2": 11544,
363
+ "5": 22730,
364
+ "4": 24790,
365
+ "1": 353,
366
+ "0": 135
367
+ },
368
+ "fixed_episode_seeds": true,
369
+ "eval_seed": 42,
370
+ "episode_seeds": [
371
+ 42,
372
+ 43,
373
+ 44,
374
+ 45,
375
+ 46,
376
+ 47,
377
+ 48,
378
+ 49,
379
+ 50,
380
+ 51,
381
+ 52,
382
+ 53,
383
+ 54,
384
+ 55,
385
+ 56,
386
+ 57,
387
+ 58,
388
+ 59,
389
+ 60,
390
+ 61
391
+ ],
392
+ "episode_indices": [
393
+ 0,
394
+ 1,
395
+ 2,
396
+ 3,
397
+ 4,
398
+ 5,
399
+ 6,
400
+ 7,
401
+ 8,
402
+ 9,
403
+ 10,
404
+ 11,
405
+ 12,
406
+ 13,
407
+ 14,
408
+ 15,
409
+ 16,
410
+ 17,
411
+ 18,
412
+ 19
413
+ ]
414
+ },
415
+ "demon_attack/latency_2": {
416
+ "latency": 2,
417
+ "num_episodes": 20,
418
+ "mean_reward": 5251.5,
419
+ "mean_length": 3533.8,
420
+ "std_reward": 755.6935556162961,
421
+ "std_length": 253.20122432563394,
422
+ "episode_rewards": [
423
+ 5410.0,
424
+ 5290.0,
425
+ 5410.0,
426
+ 4660.0,
427
+ 5780.0,
428
+ 5290.0,
429
+ 6020.0,
430
+ 5410.0,
431
+ 5410.0,
432
+ 5290.0,
433
+ 2460.0,
434
+ 5290.0,
435
+ 4330.0,
436
+ 5400.0,
437
+ 5410.0,
438
+ 5780.0,
439
+ 5290.0,
440
+ 5400.0,
441
+ 5410.0,
442
+ 6290.0
443
+ ],
444
+ "episode_lengths": [
445
+ 3600,
446
+ 3600,
447
+ 3600,
448
+ 3600,
449
+ 3521,
450
+ 3600,
451
+ 3600,
452
+ 3600,
453
+ 3600,
454
+ 3600,
455
+ 2435,
456
+ 3600,
457
+ 3600,
458
+ 3600,
459
+ 3600,
460
+ 3520,
461
+ 3600,
462
+ 3600,
463
+ 3600,
464
+ 3600
465
+ ],
466
+ "decoded_action_hist": {
467
+ "0": 5706,
468
+ "2": 8834,
469
+ "3": 7543,
470
+ "1": 31756,
471
+ "5": 9476,
472
+ "4": 7361
473
+ },
474
+ "fixed_episode_seeds": true,
475
+ "eval_seed": 42,
476
+ "episode_seeds": [
477
+ 42,
478
+ 43,
479
+ 44,
480
+ 45,
481
+ 46,
482
+ 47,
483
+ 48,
484
+ 49,
485
+ 50,
486
+ 51,
487
+ 52,
488
+ 53,
489
+ 54,
490
+ 55,
491
+ 56,
492
+ 57,
493
+ 58,
494
+ 59,
495
+ 60,
496
+ 61
497
+ ],
498
+ "episode_indices": [
499
+ 0,
500
+ 1,
501
+ 2,
502
+ 3,
503
+ 4,
504
+ 5,
505
+ 6,
506
+ 7,
507
+ 8,
508
+ 9,
509
+ 10,
510
+ 11,
511
+ 12,
512
+ 13,
513
+ 14,
514
+ 15,
515
+ 16,
516
+ 17,
517
+ 18,
518
+ 19
519
+ ]
520
+ },
521
+ "demon_attack/latency_4": {
522
+ "latency": 4,
523
+ "num_episodes": 20,
524
+ "mean_reward": 2116.5,
525
+ "mean_length": 2271.75,
526
+ "std_reward": 1479.7348580066632,
527
+ "std_length": 894.3727340991562,
528
+ "episode_rewards": [
529
+ 1900.0,
530
+ 4420.0,
531
+ 1900.0,
532
+ 3970.0,
533
+ 4760.0,
534
+ 325.0,
535
+ 1940.0,
536
+ 1640.0,
537
+ 4500.0,
538
+ 325.0,
539
+ 1760.0,
540
+ 325.0,
541
+ 3755.0,
542
+ 1480.0,
543
+ 1900.0,
544
+ 370.0,
545
+ 325.0,
546
+ 1480.0,
547
+ 1640.0,
548
+ 3615.0
549
+ ],
550
+ "episode_lengths": [
551
+ 2066,
552
+ 3355,
553
+ 2066,
554
+ 3547,
555
+ 3600,
556
+ 1129,
557
+ 2045,
558
+ 2105,
559
+ 3374,
560
+ 1129,
561
+ 2232,
562
+ 1129,
563
+ 3600,
564
+ 2083,
565
+ 2066,
566
+ 1087,
567
+ 1129,
568
+ 2083,
569
+ 2105,
570
+ 3505
571
+ ],
572
+ "decoded_action_hist": {
573
+ "0": 1154,
574
+ "3": 2344,
575
+ "1": 32941,
576
+ "4": 6228,
577
+ "5": 2521,
578
+ "2": 247
579
+ },
580
+ "fixed_episode_seeds": true,
581
+ "eval_seed": 42,
582
+ "episode_seeds": [
583
+ 42,
584
+ 43,
585
+ 44,
586
+ 45,
587
+ 46,
588
+ 47,
589
+ 48,
590
+ 49,
591
+ 50,
592
+ 51,
593
+ 52,
594
+ 53,
595
+ 54,
596
+ 55,
597
+ 56,
598
+ 57,
599
+ 58,
600
+ 59,
601
+ 60,
602
+ 61
603
+ ],
604
+ "episode_indices": [
605
+ 0,
606
+ 1,
607
+ 2,
608
+ 3,
609
+ 4,
610
+ 5,
611
+ 6,
612
+ 7,
613
+ 8,
614
+ 9,
615
+ 10,
616
+ 11,
617
+ 12,
618
+ 13,
619
+ 14,
620
+ 15,
621
+ 16,
622
+ 17,
623
+ 18,
624
+ 19
625
+ ]
626
+ },
627
+ "deadly_corridor/latency_0": {
628
+ "latency": 0,
629
+ "num_episodes": 20,
630
+ "mean_reward": 344.6194091796875,
631
+ "mean_length": 22.6,
632
+ "std_reward": 229.57653997145667,
633
+ "std_length": 8.974408058473829,
634
+ "episode_rewards": [
635
+ 79.52740478515625,
636
+ 523.2181243896484,
637
+ 578.903564453125,
638
+ 79.12068176269531,
639
+ 754.0516357421875,
640
+ 739.6824798583984,
641
+ 363.02439880371094,
642
+ 754.2959289550781,
643
+ 536.7065887451172,
644
+ 130.86195373535156,
645
+ 341.62928771972656,
646
+ 302.5681915283203,
647
+ 323.6431579589844,
648
+ 350.30462646484375,
649
+ 278.9031219482422,
650
+ 335.1339416503906,
651
+ 214.12867736816406,
652
+ 65.93748474121094,
653
+ 82.89625549316406,
654
+ 57.850677490234375
655
+ ],
656
+ "episode_lengths": [
657
+ 13,
658
+ 33,
659
+ 35,
660
+ 13,
661
+ 37,
662
+ 38,
663
+ 22,
664
+ 35,
665
+ 33,
666
+ 13,
667
+ 22,
668
+ 21,
669
+ 18,
670
+ 21,
671
+ 20,
672
+ 22,
673
+ 19,
674
+ 13,
675
+ 13,
676
+ 11
677
+ ],
678
+ "decoded_action_hist": {
679
+ "[0, 0, 0, 1, 0, 1, 0]": 39,
680
+ "[0, 1, 0, 1, 0, 1, 0]": 22,
681
+ "[0, 1, 1, 1, 0, 0, 0]": 43,
682
+ "[0, 1, 0, 1, 0, 0, 1]": 45,
683
+ "[0, 0, 0, 1, 0, 0, 0]": 148,
684
+ "[0, 1, 0, 1, 0, 0, 0]": 62,
685
+ "[0, 1, 1, 1, 0, 1, 0]": 5,
686
+ "[0, 1, 1, 1, 0, 0, 1]": 16,
687
+ "[0, 0, 1, 1, 0, 0, 0]": 55,
688
+ "[0, 0, 0, 1, 0, 0, 1]": 5,
689
+ "[1, 0, 0, 1, 0, 0, 0]": 7,
690
+ "[0, 0, 1, 1, 0, 1, 0]": 1,
691
+ "[0, 0, 1, 1, 0, 0, 1]": 4
692
+ },
693
+ "fixed_episode_seeds": true,
694
+ "eval_seed": 42,
695
+ "episode_seeds": [
696
+ 42,
697
+ 43,
698
+ 44,
699
+ 45,
700
+ 46,
701
+ 47,
702
+ 48,
703
+ 49,
704
+ 50,
705
+ 51,
706
+ 52,
707
+ 53,
708
+ 54,
709
+ 55,
710
+ 56,
711
+ 57,
712
+ 58,
713
+ 59,
714
+ 60,
715
+ 61
716
+ ],
717
+ "episode_indices": [
718
+ 0,
719
+ 1,
720
+ 2,
721
+ 3,
722
+ 4,
723
+ 5,
724
+ 6,
725
+ 7,
726
+ 8,
727
+ 9,
728
+ 10,
729
+ 11,
730
+ 12,
731
+ 13,
732
+ 14,
733
+ 15,
734
+ 16,
735
+ 17,
736
+ 18,
737
+ 19
738
+ ]
739
+ },
740
+ "deadly_corridor/latency_2": {
741
+ "latency": 2,
742
+ "num_episodes": 20,
743
+ "mean_reward": 411.0880554199219,
744
+ "mean_length": 25.3,
745
+ "std_reward": 216.71221795954787,
746
+ "std_length": 8.88313007897554,
747
+ "episode_rewards": [
748
+ 95.72293090820312,
749
+ 491.28697204589844,
750
+ 379.09344482421875,
751
+ 429.64939880371094,
752
+ 428.6657257080078,
753
+ 366.03221130371094,
754
+ 60.92265319824219,
755
+ 414.2111053466797,
756
+ 579.0622863769531,
757
+ 345.77162170410156,
758
+ 801.3818969726562,
759
+ 395.5813903808594,
760
+ 782.6199340820312,
761
+ 757.0159454345703,
762
+ 419.14808654785156,
763
+ 385.2026672363281,
764
+ 405.8075866699219,
765
+ 74.43948364257812,
766
+ 557.7950439453125,
767
+ 52.35072326660156
768
+ ],
769
+ "episode_lengths": [
770
+ 13,
771
+ 29,
772
+ 26,
773
+ 24,
774
+ 27,
775
+ 24,
776
+ 12,
777
+ 23,
778
+ 33,
779
+ 20,
780
+ 45,
781
+ 26,
782
+ 40,
783
+ 37,
784
+ 24,
785
+ 23,
786
+ 25,
787
+ 13,
788
+ 31,
789
+ 11
790
+ ],
791
+ "decoded_action_hist": {
792
+ "[0, 0, 0, 0, 0, 0, 0]": 40,
793
+ "[0, 0, 0, 1, 0, 1, 0]": 46,
794
+ "[0, 1, 1, 1, 0, 1, 0]": 14,
795
+ "[0, 1, 1, 1, 0, 0, 1]": 67,
796
+ "[0, 1, 0, 1, 0, 0, 0]": 14,
797
+ "[0, 1, 0, 1, 0, 0, 1]": 20,
798
+ "[0, 1, 1, 1, 0, 0, 0]": 36,
799
+ "[0, 0, 1, 1, 0, 0, 1]": 21,
800
+ "[0, 0, 0, 1, 0, 0, 0]": 38,
801
+ "[1, 0, 0, 1, 0, 0, 0]": 30,
802
+ "[1, 0, 1, 1, 0, 0, 0]": 27,
803
+ "[0, 0, 1, 1, 0, 0, 0]": 96,
804
+ "[0, 0, 0, 1, 0, 0, 1]": 8,
805
+ "[0, 0, 1, 1, 0, 1, 0]": 13,
806
+ "[1, 0, 1, 1, 0, 0, 1]": 7,
807
+ "[0, 1, 1, 1, 0, 1, 1]": 2,
808
+ "[1, 0, 0, 1, 0, 1, 0]": 19,
809
+ "[0, 0, 0, 1, 1, 1, 0]": 6,
810
+ "[1, 0, 1, 1, 0, 1, 0]": 1,
811
+ "[0, 0, 0, 0, 1, 1, 0]": 1
812
+ },
813
+ "fixed_episode_seeds": true,
814
+ "eval_seed": 42,
815
+ "episode_seeds": [
816
+ 42,
817
+ 43,
818
+ 44,
819
+ 45,
820
+ 46,
821
+ 47,
822
+ 48,
823
+ 49,
824
+ 50,
825
+ 51,
826
+ 52,
827
+ 53,
828
+ 54,
829
+ 55,
830
+ 56,
831
+ 57,
832
+ 58,
833
+ 59,
834
+ 60,
835
+ 61
836
+ ],
837
+ "episode_indices": [
838
+ 0,
839
+ 1,
840
+ 2,
841
+ 3,
842
+ 4,
843
+ 5,
844
+ 6,
845
+ 7,
846
+ 8,
847
+ 9,
848
+ 10,
849
+ 11,
850
+ 12,
851
+ 13,
852
+ 14,
853
+ 15,
854
+ 16,
855
+ 17,
856
+ 18,
857
+ 19
858
+ ]
859
+ },
860
+ "deadly_corridor/latency_4": {
861
+ "latency": 4,
862
+ "num_episodes": 20,
863
+ "mean_reward": 193.83389129638672,
864
+ "mean_length": 18.6,
865
+ "std_reward": 186.45049743335858,
866
+ "std_length": 6.374950980203691,
867
+ "episode_rewards": [
868
+ 71.78205871582031,
869
+ 18.258163452148438,
870
+ 510.49853515625,
871
+ 41.72627258300781,
872
+ 587.3656005859375,
873
+ 205.68914794921875,
874
+ 52.63993835449219,
875
+ 429.2797088623047,
876
+ 15.178573608398438,
877
+ 61.64399719238281,
878
+ 124.24365234375,
879
+ 180.68826293945312,
880
+ 355.9683074951172,
881
+ 134.1019744873047,
882
+ 378.20603942871094,
883
+ 493.99058532714844,
884
+ 125.47468566894531,
885
+ 2.539886474609375,
886
+ 66.77407836914062,
887
+ 20.62835693359375
888
+ ],
889
+ "episode_lengths": [
890
+ 13,
891
+ 12,
892
+ 27,
893
+ 13,
894
+ 32,
895
+ 20,
896
+ 12,
897
+ 27,
898
+ 12,
899
+ 13,
900
+ 20,
901
+ 21,
902
+ 22,
903
+ 21,
904
+ 22,
905
+ 29,
906
+ 19,
907
+ 13,
908
+ 13,
909
+ 11
910
+ ],
911
+ "decoded_action_hist": {
912
+ "[0, 0, 0, 0, 0, 0, 0]": 80,
913
+ "[0, 0, 1, 1, 0, 0, 0]": 43,
914
+ "[1, 0, 1, 1, 0, 0, 0]": 76,
915
+ "[0, 0, 0, 1, 0, 1, 0]": 9,
916
+ "[0, 1, 1, 1, 0, 0, 0]": 23,
917
+ "[1, 0, 1, 1, 0, 0, 1]": 26,
918
+ "[0, 1, 1, 1, 0, 0, 1]": 42,
919
+ "[0, 0, 1, 1, 0, 0, 1]": 18,
920
+ "[0, 0, 0, 1, 0, 0, 0]": 6,
921
+ "[0, 0, 1, 1, 0, 1, 0]": 17,
922
+ "[1, 0, 1, 1, 0, 1, 0]": 13,
923
+ "[1, 0, 0, 1, 0, 1, 0]": 9,
924
+ "[1, 0, 0, 1, 0, 0, 0]": 8,
925
+ "[1, 1, 1, 1, 0, 0, 0]": 1,
926
+ "[0, 0, 1, 1, 1, 0, 0]": 1
927
+ },
928
+ "fixed_episode_seeds": true,
929
+ "eval_seed": 42,
930
+ "episode_seeds": [
931
+ 42,
932
+ 43,
933
+ 44,
934
+ 45,
935
+ 46,
936
+ 47,
937
+ 48,
938
+ 49,
939
+ 50,
940
+ 51,
941
+ 52,
942
+ 53,
943
+ 54,
944
+ 55,
945
+ 56,
946
+ 57,
947
+ 58,
948
+ 59,
949
+ 60,
950
+ 61
951
+ ],
952
+ "episode_indices": [
953
+ 0,
954
+ 1,
955
+ 2,
956
+ 3,
957
+ 4,
958
+ 5,
959
+ 6,
960
+ 7,
961
+ 8,
962
+ 9,
963
+ 10,
964
+ 11,
965
+ 12,
966
+ 13,
967
+ 14,
968
+ 15,
969
+ 16,
970
+ 17,
971
+ 18,
972
+ 19
973
+ ]
974
+ }
975
+ },
976
+ "aggregate": {
977
+ "stage": "post_train",
978
+ "step": 15000,
979
+ "task": "cross_task",
980
+ "model_alias": "openvla",
981
+ "fixed_episode_seeds": true,
982
+ "eval_seed": 42,
983
+ "total_episodes": 180,
984
+ "mean_reward": 1841.5034839884502,
985
+ "mean_length": 2096.2555555555555,
986
+ "std_reward": 2488.2710614030866,
987
+ "std_length": 1620.56384948169,
988
+ "task_count": 3,
989
+ "macro_mean_reward": 1841.5034839884502,
990
+ "macro_mean_length": 2096.2555555555555,
991
+ "distributed_eval": false
992
+ }
993
+ }
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/hydra/train_starvla_hydra.log ADDED
The diff for this file is too large to render. See raw diff
 
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/summary.jsonl ADDED
@@ -0,0 +1 @@
 
 
1
+ {"steps": 15000}
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/debug-internal.log ADDED
The diff for this file is too large to render. See raw diff
 
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/debug.log ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_setup.py:_flush():81] Current SDK version is 0.26.1
2
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_setup.py:_flush():81] Configure stats pid to 517811
3
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_setup.py:_flush():81] Loading settings from environment variables
4
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_init.py:setup_run_log_directory():723] Logging user logs to /mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_105406-7ujo2tgu/logs/debug.log
5
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_init.py:setup_run_log_directory():724] Logging internal logs to /mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_105406-7ujo2tgu/logs/debug-internal.log
6
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_init.py:init():850] calling init triggers
7
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_init.py:init():855] wandb.init called with sweep_config: {}
8
+ config: {'_wandb': {}}
9
+ 2026-09-23 10:54:06,379 INFO MainThread:517811 [wandb_init.py:init():898] starting backend
10
+ 2026-09-23 10:54:06,627 INFO MainThread:517811 [wandb_init.py:init():913] sending inform_init request
11
+ 2026-09-23 10:54:08,005 INFO MainThread:517811 [wandb_init.py:init():918] backend started and connected
12
+ 2026-09-23 10:54:08,006 INFO MainThread:517811 [wandb_init.py:init():988] updated telemetry
13
+ 2026-09-23 10:54:08,010 INFO MainThread:517811 [wandb_init.py:init():1011] communicating run to backend with 90.0 second timeout
14
+ 2026-09-23 10:54:08,585 INFO MainThread:517811 [wandb_init.py:init():1056] starting run threads in backend
15
+ 2026-09-23 10:54:08,730 INFO MainThread:517811 [wandb_run.py:_console_start():2554] atexit reg
16
+ 2026-09-23 10:54:08,730 INFO MainThread:517811 [wandb_run.py:_redirect():2403] redirect: wrap_raw
17
+ 2026-09-23 10:54:08,731 INFO MainThread:517811 [wandb_run.py:_redirect():2472] Wrapping output streams.
18
+ 2026-09-23 10:54:08,731 INFO MainThread:517811 [wandb_run.py:_redirect():2495] Redirects installed.
19
+ 2026-09-23 10:54:08,733 INFO MainThread:517811 [wandb_init.py:init():1094] run started, returning control to user process
20
+ 2026-09-23 10:54:08,734 INFO MainThread:517811 [wandb_run.py:_config_callback():1415] config_cb None None {'micro_batch': 32, 'gradient_accumulation_steps': 8, 'global_batch': 256, 'training_latency_condition': 'rtx3090_profile'}
21
+ 2026-09-24 17:30:32,557 INFO MainThread:517811 [wandb_run.py:_finish():2321] finishing run talha1503/starVLA_rl_games/7ujo2tgu
22
+ 2026-09-24 17:30:32,558 INFO MainThread:517811 [wandb_run.py:_atexit_cleanup():2520] got exitcode: 0
23
+ 2026-09-24 17:30:32,558 INFO MainThread:517811 [wandb_run.py:_restore():2502] restore
24
+ 2026-09-24 17:30:32,558 INFO MainThread:517811 [wandb_run.py:_restore():2508] restore done
25
+ 2026-09-24 17:30:38,863 INFO MainThread:517811 [wandb_run.py:_footer_sync_info():3889] logging synced files
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/files/wandb-summary.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"_wandb":{"runtime":0},"_runtime":0}
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug-core.log ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-09-23T06:46:45.054599604Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp0yezjqan/port-2322819.txt","pid":2322819,"detached":false,"idle-timeout":600000000000,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
2
+ {"time":"2026-09-23T06:46:45.054899633Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":2322819}
3
+ {"time":"2026-09-23T06:46:45.054893977Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-2322819-2701894-2375320025/socket","Net":"unix"}}
4
+ {"time":"2026-09-23T06:46:45.219212209Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
5
+ {"time":"2026-09-23T06:46:45.22491556Z","level":"INFO","msg":"handleInformInit: received","streamId":"r8358sud","id":"1(@)"}
6
+ {"time":"2026-09-23T06:46:45.830481405Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"r8358sud","id":"1(@)"}
7
+ {"time":"2026-09-23T06:46:47.021707351Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
8
+ {"time":"2026-09-23T06:46:47.021778217Z","level":"INFO","msg":"server is shutting down"}
9
+ {"time":"2026-09-23T06:46:47.02179905Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
10
+ {"time":"2026-09-23T06:46:47.021865099Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
11
+ {"time":"2026-09-23T06:46:47.021884377Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-2322819-2701894-2375320025/socket","Net":"unix"}}
12
+ {"time":"2026-09-23T06:46:47.023715559Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
13
+ {"time":"2026-09-23T06:46:47.023738152Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
14
+ {"time":"2026-09-23T06:46:47.023745449Z","level":"INFO","msg":"server is closed"}
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug-internal.log ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-09-23T06:46:45.225033265Z","level":"INFO","msg":"wandb-core"}
2
+ {"time":"2026-09-23T06:46:45.22529674Z","level":"INFO","msg":"stream: starting","core version":"0.26.1"}
3
+ {"time":"2026-09-23T06:46:45.830296549Z","level":"INFO","msg":"stream: created new stream","id":"r8358sud"}
4
+ {"time":"2026-09-23T06:46:45.830372055Z","level":"INFO","msg":"handler: started"}
5
+ {"time":"2026-09-23T06:46:45.830473435Z","level":"INFO","msg":"stream: started"}
6
+ {"time":"2026-09-23T06:46:45.830493427Z","level":"INFO","msg":"writer: started","stream_id":"r8358sud"}
7
+ {"time":"2026-09-23T06:46:45.830511339Z","level":"INFO","msg":"sender: started"}
8
+ {"time":"2026-09-23T06:46:46.143990549Z","level":"ERROR","msg":"api: HTTP error","status":403,"method":"POST","url":"https://api.wandb.ai/graphql"}
9
+ {"time":"2026-09-23T06:46:46.144079893Z","level":"ERROR","msg":"runupserter: failed to init run","error":"returned error 403: {\"data\":{\"upsertBucket\":null},\"errors\":[{\"message\":\"permission denied\",\"path\":[\"upsertBucket\"],\"extensions\":{\"code\":\"PERMISSION_ERROR\"}}]}"}
10
+ {"time":"2026-09-23T06:46:47.022186839Z","level":"ERROR","msg":"sender: uploadConfigFile: runhandle: run not yet initialized"}
11
+ {"time":"2026-09-23T06:46:47.022436007Z","level":"ERROR","msg":"runfiles: runhandle: run not yet initialized"}
12
+ {"time":"2026-09-23T06:46:47.022515322Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
13
+ {"time":"2026-09-23T06:46:47.022524274Z","level":"INFO","msg":"stream: finishing up"}
14
+ {"time":"2026-09-23T06:46:47.022530305Z","level":"INFO","msg":"handler: closed"}
15
+ {"time":"2026-09-23T06:46:47.022555383Z","level":"INFO","msg":"sender: closed"}
16
+ {"time":"2026-09-23T06:46:47.022559583Z","level":"INFO","msg":"stream: all finished"}
17
+ {"time":"2026-09-23T06:46:47.023616018Z","level":"ERROR","msg":"stream: runURL: runhandle: run not yet initialized"}
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug.log ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_setup.py:_flush():81] Current SDK version is 0.26.1
2
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_setup.py:_flush():81] Configure stats pid to 2322819
3
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_setup.py:_flush():81] Loading settings from environment variables
4
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_init.py:setup_run_log_directory():723] Logging user logs to /mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug.log
5
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_init.py:setup_run_log_directory():724] Logging internal logs to /mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/logs/debug-internal.log
6
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_init.py:init():850] calling init triggers
7
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_init.py:init():855] wandb.init called with sweep_config: {}
8
+ config: {'_wandb': {}}
9
+ 2026-09-23 06:46:44,966 INFO MainThread:2322819 [wandb_init.py:init():898] starting backend
10
+ 2026-09-23 06:46:45,219 INFO MainThread:2322819 [wandb_init.py:init():913] sending inform_init request
11
+ 2026-09-23 06:46:45,831 INFO MainThread:2322819 [wandb_init.py:init():918] backend started and connected
12
+ 2026-09-23 06:46:45,832 INFO MainThread:2322819 [wandb_init.py:init():988] updated telemetry
13
+ 2026-09-23 06:46:45,836 INFO MainThread:2322819 [wandb_init.py:init():1011] communicating run to backend with 90.0 second timeout
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_064644-r8358sud/run-r8358sud.wandb ADDED
Binary file (616 Bytes). View file
 
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/config.yaml ADDED
@@ -0,0 +1,417 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ _wandb:
2
+ value:
3
+ cli_version: 0.26.1
4
+ e:
5
+ m2yr6468uhdnlzudseussxn2vn8jrqko:
6
+ args:
7
+ - --config-name
8
+ - train
9
+ - model=openvla
10
+ - env=cross_task
11
+ - init=bridge
12
+ - mode=cross_task
13
+ - ++framework.qwenvl.base_vlm=/mnt/local/talha/playground/Pretrained_models/Qwen3-VL-4B-Instruct
14
+ - ++framework.qwenvl.attn_implementation=flash_attention_2
15
+ - ++framework.qwenvl.enable_gradient_checkpointing=true
16
+ - ++framework.action_model.state_dim=7
17
+ - ++framework.action_model.loss_type=discrete_ce
18
+ - ++framework.action_model.action_horizon=1
19
+ - ++framework.action_model.future_action_window_size=0
20
+ - ++framework.action_model.past_action_window_size=0
21
+ - ++framework.action_model.action_dim=7
22
+ - ++framework.action_model.action_env_dim=7
23
+ - ++framework.action_model.state_encoding=discretized_text
24
+ - ++framework.action_model.action_layout=bridge_cross_task_7
25
+ - ++framework.kv_memory.enabled=false
26
+ - ++framework.kv_memory.window=4
27
+ - ++framework.kv_memory.rollout_len=8
28
+ - ++framework.kv_memory.packed_train=false
29
+ - ++framework.kv_memory.rebased_sink=true
30
+ - ++framework.name=QwenOFT
31
+ - ++datasets.vla_data.dataset_py=lerobot_datasets
32
+ - ++datasets.vla_data.include_state=false
33
+ - ++datasets.vla_data.data_root_dir=playground/Datasets/rl_games
34
+ - ++datasets.vla_data.data_mix=cross_task_train
35
+ - ++datasets.vla_data.eval_data_mix=null
36
+ - ++datasets.vla_data.custom_mixtures_path=null
37
+ - ++datasets.vla_data.action_type=discrete
38
+ - ++datasets.vla_data.sequential_step_sampling=true
39
+ - ++datasets.vla_data.eval_sequential_step_sampling=null
40
+ - ++datasets.vla_data.num_workers=8
41
+ - ++datasets.vla_data.eval_num_workers=8
42
+ - ++datasets.vla_data.prefetch_factor=4
43
+ - ++datasets.vla_data.persistent_workers=true
44
+ - ++datasets.vla_data.pin_memory=true
45
+ - ++datasets.vla_data.shuffle=true
46
+ - ++datasets.vla_data.action_balance.enabled=false
47
+ - ++datasets.vla_data.action_balance.strategy=balanced_epoch
48
+ - ++datasets.vla_data.action_balance.action_key=action_id
49
+ - ++datasets.vla_data.action_balance.target_flap_fraction=0.3
50
+ - ++datasets.vla_data.action_balance.noop_id=0
51
+ - ++datasets.vla_data.action_balance.flap_id=1
52
+ - ++datasets.vla_data.latency_curriculum.enabled=false
53
+ - ++datasets.vla_data.latency_curriculum.strategy=exclusive
54
+ - ++datasets.vla_data.latency_curriculum.latencies=null
55
+ - ++datasets.vla_data.latency_curriculum.phase_steps=null
56
+ - ++datasets.vla_data.latency_curriculum.phase_distributions=null
57
+ - ++datasets.vla_data.latency_curriculum.new_latency_passes=1.0
58
+ - ++datasets.vla_data.latency_curriculum.replay_passes=0.25
59
+ - ++datasets.vla_data.latency_curriculum.target_total_passes=2.0
60
+ - ++datasets.vla_data.latency_curriculum.final_equalization=true
61
+ - ++datasets.vla_data.latency_curriculum.step_budget_mode=auto
62
+ - ++datasets.vla_data.latency_curriculum.eval_at_phase_end=false
63
+ - ++datasets.vla_data.latency_curriculum.save_at_phase_end=false
64
+ - ++datasets.vla_data.latency_curriculum.computed_plan=null
65
+ - ++datasets.vla_data.per_device_batch_size=32
66
+ - ++datasets.vla_data.load_all_data_for_training=true
67
+ - ++datasets.vla_data.num_obs_frames=1
68
+ - ++datasets.vla_data.image_mode=single
69
+ - ++datasets.vla_data.prompt_mode=raw
70
+ - ++datasets.vla_data.stitch_grid=[2,2]
71
+ - ++datasets.vla_data.obs_image_size=null
72
+ - ++datasets.vla_data.video_backend=torchvision_av
73
+ - ++dataset.source_hf=
74
+ - ++dataset.config_name=null
75
+ - ++dataset.source_subdir=null
76
+ - ++dataset.converted_name=cross_task_train
77
+ - ++dataset.single_source_hf=
78
+ - ++dataset.mixed_source_hf=
79
+ - ++dataset.single_converted_name=flappy_train
80
+ - ++dataset.mixed_converted_name=flappy_mixed_latency_train
81
+ - ++dataset.single_latency_filter=null
82
+ - ++dataset.mixed_latency_filter=null
83
+ - ++dataset.force_download=false
84
+ - ++dataset.setup_force=false
85
+ - ++dataset.skip_verification=false
86
+ - ++dataset.target_latency_unit=observation_steps
87
+ - ++dataset.verify_rows=200
88
+ - ++dataset.max_episodes=null
89
+ - ++dataset.episodes_per_latency=null
90
+ - ++dataset.latency_filter=null
91
+ - ++dataset.debug_subset.enabled=false
92
+ - ++dataset.debug_subset.max_episodes=5
93
+ - ++dataset.debug_subset.suffix=debug
94
+ - ++base_model.repo_id=Qwen/Qwen3-VL-4B-Instruct
95
+ - ++initialization.checkpoint_local_dir=playground/Pretrained_models/Qwen3VL-OFT-Bridge-RT-1
96
+ - ++initialization.checkpoint_hf_repo_id=StarVLA/Qwen3VL-OFT-Bridge-RT-1
97
+ - ++initialization.checkpoint_filename=checkpoints/steps_5000_pytorch_model.pt
98
+ - ++trainer.max_train_steps=15000
99
+ - ++trainer.num_warmup_steps=0
100
+ - ++trainer.save_interval=15000
101
+ - ++trainer.eval_interval=15000
102
+ - ++trainer.eval_num_batches=100
103
+ - ++trainer.per_latency_eval_num_batches=5
104
+ - ++trainer.eval_action_classification=true
105
+ - ++trainer.eval_action_classification_interval=null
106
+ - ++trainer.cc_f1_tolerance=1
107
+ - ++trainer.learning_rate.base=2e-05
108
+ - ++trainer.learning_rate.qwen_vl_interface=1e-05
109
+ - ++trainer.learning_rate.action_model=0.0001
110
+ - ++trainer.lr_scheduler_type=cosine_with_min_lr
111
+ - ++trainer.scheduler_specific_kwargs.min_lr=1e-06
112
+ - ++trainer.freeze_modules=
113
+ - ++trainer.freeze_vit=false
114
+ - ++trainer.freeze_tied_embedding=false
115
+ - ++trainer.freeze_llm_layers=[]
116
+ - ++trainer.loss_scale.vla=1.0
117
+ - ++trainer.loss_scale.vlm=0.1
118
+ - ++trainer.max_grad_norm=1.0
119
+ - ++trainer.weight_decay=0.0
120
+ - ++trainer.logging_frequency=1
121
+ - ++trainer.profile_timing.enabled=false
122
+ - ++trainer.profile_timing.log_interval=10
123
+ - ++trainer.gradient_clipping=1.0
124
+ - ++trainer.gradient_accumulation_steps=8
125
+ - ++trainer.distributed_backend=none
126
+ - ++trainer.is_resume=false
127
+ - ++trainer.pretrained_checkpoint=null
128
+ - ++trainer.resume_step=0
129
+ - ++trainer.reload_modules=null
130
+ - ++trainer.optimizer.name=AdamW
131
+ - ++trainer.optimizer.betas=[0.9,0.95]
132
+ - ++trainer.optimizer.eps=1e-08
133
+ - ++trainer.optimizer.weight_decay=1e-08
134
+ - ++trainer.optimizer.fused=true
135
+ - ++trainer.save_format=pt
136
+ - ++workspace_dir=/mnt/local/talha
137
+ - ++run_root_dir=results/Checkpoints
138
+ - ++seed=42
139
+ - ++auth.env_file=null
140
+ - ++auth.hf_token_env=HF_TOKEN
141
+ - ++auth.wandb_api_key_env=WANDB_API_KEY
142
+ - ++paths.run_root_dir=/mnt/local/talha/results/Checkpoints
143
+ - ++paths.dataset_local_dir=/mnt/local/talha/playground/Datasets/rl_games
144
+ - ++paths.dataset_cache_dir=null
145
+ - ++paths.base_model_dir=/mnt/local/talha/playground/Pretrained_models/Qwen3-VL-4B-Instruct
146
+ - ++paths.accelerate_config=starVLA/config/deepseeds/deepspeed_zero2.yaml
147
+ - ++rl_games.model_alias=openvla
148
+ - ++rl_games.task=cross_task
149
+ - ++rl_games.env_eval.image_size=224
150
+ - ++rl_games.env_eval.frameskip=1
151
+ - ++rl_games.env_eval.deadly.action_layout=multibinary_7
152
+ - ++rl_games.env_eval.deadly.multibinary_threshold=0.0
153
+ - ++rl_games.env_eval.enabled=true
154
+ - ++rl_games.env_eval.eval_backend=eval_core
155
+ - ++rl_games.env_eval.distributed_mode=none
156
+ - ++rl_games.env_eval.vectorized.enabled=false
157
+ - ++rl_games.env_eval.vectorized.batch_size=1
158
+ - ++rl_games.env_eval.latency.prompt_map_path=null
159
+ - ++rl_games.env_eval.latency.mode=mixed
160
+ - ++rl_games.env_eval.latency.values=[0,1,2,3,4,5]
161
+ - ++rl_games.env_eval.mid_train.enabled=true
162
+ - ++rl_games.env_eval.mid_train.interval_steps=1000
163
+ - ++rl_games.env_eval.mid_train.latencies=[0,1,2,3,4,5]
164
+ - ++rl_games.env_eval.mid_train.num_episodes=5
165
+ - ++rl_games.env_eval.mid_train.max_steps_per_episode=3600
166
+ - ++rl_games.env_eval.post_train.enabled=true
167
+ - ++rl_games.env_eval.post_train.latencies=[0,1,2,3,4]
168
+ - ++rl_games.env_eval.post_train.latencies_from_prompt_map=false
169
+ - ++rl_games.env_eval.post_train.num_episodes=5
170
+ - ++rl_games.env_eval.post_train.max_steps_per_episode=3600
171
+ - ++rl_games.env_eval.asterix.action_layout=factorized_6
172
+ - ++rl_games.cross_task.train_tasks=[{name:flappy,role:A,train_source_hf:latency-sensitive-bench/flappy_200ep,prompt_source_hf:latency-sensitive-bench/flappy_200ep,train_source_subdir:flappy_fix_latency_0_200ep,prompt_source_subdir:flappy_fix_latency_0_200ep,converted_name:flappy_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0},{name:demon_attack,role:B,train_source_hf:latency-sensitive-bench/demon_attack_200ep,prompt_source_hf:latency-sensitive-bench/demon_attack_200ep,train_source_subdir:demon_attack_fix_latency_0_200ep,prompt_source_subdir:demon_attack_fix_latency_0_200ep,converted_name:demon_attack_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0},{name:deadly_corridor,role:C,train_source_hf:latency-sensitive-bench/deadly_1000ep,prompt_source_hf:latency-sensitive-bench/deadly_1000ep,train_source_subdir:deadly_corridor_fix_latency_0_1000ep,prompt_source_subdir:deadly_corridor_fix_latency_0_1000ep,converted_name:deadly_corridor_mixed_024_1000ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:1000,action_layout:multibinary_7,weight:1.0},{name:asterix,role:D,train_source_hf:latency-sensitive-bench/asterix_200ep,prompt_source_hf:latency-sensitive-bench/asterix_200ep,train_source_subdir:asterix_fixed_latency_0_200ep_7k2steps,prompt_source_subdir:asterix_fixed_latency_0_200ep_7k2steps,converted_name:asterix_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,action_layout:factorized_6,weight:1.0},{name:atlantis,role:E,train_source_hf:latency-sensitive-bench/atlantis_200ep,prompt_source_hf:latency-sensitive-bench/atlantis_200ep,train_source_subdir:atlantis_fixed_latency_0_200ep_7k2steps,prompt_source_subdir:atlantis_fixed_latency_0_200ep_7k2steps,converted_name:atlantis_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0},{name:defend_the_line,role:F,train_source_hf:latency-sensitive-bench/defend_the_line_200ep,prompt_source_hf:latency-sensitive-bench/defend_the_line_200ep,train_source_subdir:defend_the_line_fixed_latency_0_200ep_7k2steps,prompt_source_subdir:defend_the_line_fixed_latency_0_200ep_7k2steps,converted_name:defend_the_line_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0}]
173
+ - ++rl_games.cross_task.loss_by_task.flappy=discrete_ce
174
+ - ++rl_games.cross_task.loss_by_task.demon_attack=discrete_ce
175
+ - ++rl_games.cross_task.loss_by_task.defend_the_line=discrete_ce
176
+ - ++rl_games.cross_task.loss_by_task.deadly_corridor=multibinary_bce
177
+ - ++rl_games.cross_task.loss_by_task.asterix=asterix_factorized_ce
178
+ - ++rl_games.cross_task.loss_by_task.atlantis=discrete_ce
179
+ - ++rl_games.cross_task.loss_weight_by_task.flappy=1.0
180
+ - ++rl_games.cross_task.loss_weight_by_task.demon_attack=1.0
181
+ - ++rl_games.cross_task.loss_weight_by_task.defend_the_line=1.0
182
+ - ++rl_games.cross_task.loss_weight_by_task.deadly_corridor=1.0
183
+ - ++rl_games.cross_task.loss_weight_by_task.asterix=1.0
184
+ - ++rl_games.cross_task.loss_weight_by_task.atlantis=1.0
185
+ - ++rl_games.cross_task.eval_tasks.flappy.prompt_map_path=null
186
+ - ++rl_games.cross_task.eval_tasks.flappy.frameskip=1
187
+ - ++rl_games.cross_task.eval_tasks.flappy.image_size=224
188
+ - '++rl_games.cross_task.eval_tasks.flappy.task_description=''You are playing Flappy Bird. Pass through the pipe gaps and stay alive. Choose the action: NOOP, FLAP.'''
189
+ - ++rl_games.cross_task.eval_tasks.flappy.mid_train.enabled=false
190
+ - ++rl_games.cross_task.eval_tasks.flappy.mid_train.latencies=[0,2,4]
191
+ - ++rl_games.cross_task.eval_tasks.flappy.mid_train.num_episodes=5
192
+ - ++rl_games.cross_task.eval_tasks.flappy.mid_train.max_steps_per_episode=3600
193
+ - ++rl_games.cross_task.eval_tasks.flappy.post_train.enabled=true
194
+ - ++rl_games.cross_task.eval_tasks.flappy.post_train.latencies=[0,2,4]
195
+ - ++rl_games.cross_task.eval_tasks.flappy.post_train.num_episodes=20
196
+ - ++rl_games.cross_task.eval_tasks.flappy.post_train.max_steps_per_episode=3600
197
+ - ++rl_games.cross_task.eval_tasks.demon_attack.prompt_map_path=null
198
+ - ++rl_games.cross_task.eval_tasks.demon_attack.frameskip=4
199
+ - ++rl_games.cross_task.eval_tasks.demon_attack.image_size=224
200
+ - '++rl_games.cross_task.eval_tasks.demon_attack.task_description=''You are playing Demon Attack from a single game image. Choose exactly one action from: NOOP, FIRE, RIGHT, LEFT, RIGHTFIRE, LEFTFIRE.'''
201
+ - ++rl_games.cross_task.eval_tasks.demon_attack.mid_train.enabled=false
202
+ - ++rl_games.cross_task.eval_tasks.demon_attack.mid_train.latencies=[0,2,4]
203
+ - ++rl_games.cross_task.eval_tasks.demon_attack.mid_train.num_episodes=5
204
+ - ++rl_games.cross_task.eval_tasks.demon_attack.mid_train.max_steps_per_episode=3600
205
+ - ++rl_games.cross_task.eval_tasks.demon_attack.post_train.enabled=true
206
+ - ++rl_games.cross_task.eval_tasks.demon_attack.post_train.latencies=[0,2,4]
207
+ - ++rl_games.cross_task.eval_tasks.demon_attack.post_train.num_episodes=20
208
+ - ++rl_games.cross_task.eval_tasks.demon_attack.post_train.max_steps_per_episode=3600
209
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.prompt_map_path=null
210
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.frameskip=4
211
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.image_size=224
212
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.task_description=null
213
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.enabled=true
214
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.latencies=null
215
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.num_episodes=null
216
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.max_steps_per_episode=null
217
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.post_train.enabled=true
218
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.post_train.latencies=null
219
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.post_train.num_episodes=null
220
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.post_train.max_steps_per_episode=null
221
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.prompt_map_path=null
222
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.frameskip=4
223
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.image_size=224
224
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.task_description='You are playing Deadly Corridor in VizDoom. Choose actions from MOVE_FORWARD, MOVE_BACKWARD, MOVE_LEFT, MOVE_RIGHT, TURN_LEFT, TURN_RIGHT, ATTACK.'
225
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.enabled=false
226
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.latencies=[0,2,4]
227
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.num_episodes=5
228
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.max_steps_per_episode=3600
229
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.enabled=true
230
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.latencies=[0,2,4]
231
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.num_episodes=20
232
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.max_steps_per_episode=3600
233
+ - ++rl_games.cross_task.eval_tasks.asterix.prompt_map_path=null
234
+ - ++rl_games.cross_task.eval_tasks.asterix.frameskip=4
235
+ - ++rl_games.cross_task.eval_tasks.asterix.image_size=224
236
+ - ++rl_games.cross_task.eval_tasks.asterix.task_description=null
237
+ - ++rl_games.cross_task.eval_tasks.asterix.mid_train.enabled=true
238
+ - ++rl_games.cross_task.eval_tasks.asterix.mid_train.latencies=null
239
+ - ++rl_games.cross_task.eval_tasks.asterix.mid_train.num_episodes=null
240
+ - ++rl_games.cross_task.eval_tasks.asterix.mid_train.max_steps_per_episode=null
241
+ - ++rl_games.cross_task.eval_tasks.asterix.post_train.enabled=true
242
+ - ++rl_games.cross_task.eval_tasks.asterix.post_train.latencies=null
243
+ - ++rl_games.cross_task.eval_tasks.asterix.post_train.num_episodes=null
244
+ - ++rl_games.cross_task.eval_tasks.asterix.post_train.max_steps_per_episode=null
245
+ - ++rl_games.cross_task.eval_tasks.atlantis.prompt_map_path=null
246
+ - ++rl_games.cross_task.eval_tasks.atlantis.frameskip=4
247
+ - ++rl_games.cross_task.eval_tasks.atlantis.image_size=224
248
+ - ++rl_games.cross_task.eval_tasks.atlantis.task_description=null
249
+ - ++rl_games.cross_task.eval_tasks.atlantis.mid_train.enabled=true
250
+ - ++rl_games.cross_task.eval_tasks.atlantis.mid_train.latencies=null
251
+ - ++rl_games.cross_task.eval_tasks.atlantis.mid_train.num_episodes=null
252
+ - ++rl_games.cross_task.eval_tasks.atlantis.mid_train.max_steps_per_episode=null
253
+ - ++rl_games.cross_task.eval_tasks.atlantis.post_train.enabled=true
254
+ - ++rl_games.cross_task.eval_tasks.atlantis.post_train.latencies=null
255
+ - ++rl_games.cross_task.eval_tasks.atlantis.post_train.num_episodes=null
256
+ - ++rl_games.cross_task.eval_tasks.atlantis.post_train.max_steps_per_episode=null
257
+ - ++rl_games.initialization_mode=bridge
258
+ - ++rl_games.action_carrier=bridge
259
+ - ++bridge_base_model.repo_id.openvla=Qwen/Qwen3-VL-4B-Instruct
260
+ - ++bridge_base_model.repo_id.pi0=StarVLA/Qwen2.5-VL-3B-Instruct-Action
261
+ - ++bridge_base_model.repo_id.pi05=Qwen/Qwen3-VL-4B-Instruct
262
+ - ++bridge_base_model.repo_id.gr00t=Qwen/Qwen3-VL-4B-Instruct
263
+ - ++bridge_base_model.local_dir.openvla=playground/Pretrained_models/Qwen3-VL-4B-Instruct
264
+ - ++bridge_base_model.local_dir.pi0=playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action
265
+ - ++bridge_base_model.local_dir.pi05=playground/Pretrained_models/Qwen3-VL-4B-Instruct
266
+ - ++bridge_base_model.local_dir.gr00t=playground/Pretrained_models/Qwen3-VL-4B-Instruct
267
+ - ++checkpoint.load=auto
268
+ - ++checkpoint.hf_repo_id=latency-sensitive-bench/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1
269
+ - ++checkpoint.save_best_model=false
270
+ - ++checkpoint.save_final_model=true
271
+ - ++checkpoint.save_pt_file=false
272
+ - ++checkpoint.save_training_state=true
273
+ - ++checkpoint.save_safetensors_file=false
274
+ - ++checkpoint.local.keep_last_n=1
275
+ - ++checkpoint.sync.enabled=true
276
+ - ++checkpoint.sync.repo_id=latency-sensitive-bench/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1
277
+ - ++checkpoint.sync.keep_last_n=0
278
+ - ++checkpoint.sync.sync_every_n_checkpoints=1
279
+ - ++checkpoint.sync.resume_policy=local_latest
280
+ - ++run_id=openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1
281
+ - ++output_dir=null
282
+ - ++config_yaml=null
283
+ - ++is_debug=false
284
+ - ++version_id=0.21
285
+ - ++wandb_tags=[final_vla_profiling_training,demon_attack,openvla,rtx3090_profile]
286
+ - ++training_latency_condition=rtx3090_profile
287
+ - ++wandb_group=final_vla_profiling_training
288
+ - ++run_root_dir=/mnt/local/talha/results/Checkpoints
289
+ - ++trainer.is_resume=false
290
+ - ++trainer.pretrained_checkpoint=/mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/_initialization/StarVLA_Qwen3VL-OFT-Bridge-RT-1/checkpoints/steps_5000_pytorch_model.pt
291
+ - ++trainer.resume_step=0
292
+ - ++datasets.vla_data.data_root_dir=/mnt/local/talha/playground/Datasets/rl_games
293
+ - ++datasets.vla_data.data_mix=cross__asterix_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__atlantis_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__deadly_corridor_mixed_024_1000ep_cross_train__lat0_2_4__1000ep__bridge__defend_the_line_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__demon_attack_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__flappy_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge
294
+ - ++datasets.vla_data.eval_data_mix=cross__asterix_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__atlantis_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__deadly_corridor_mixed_024_1000ep_cross_train__lat0_2_4__1000ep__bridge__defend_the_line_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__demon_attack_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__flappy_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__val
295
+ - ++datasets.vla_data.custom_mixtures_path=/mnt/local/talha/playground/Datasets/rl_games/_generated_mixtures/cross_09b5517c91a12cca.json
296
+ - ++framework.qwenvl.base_vlm=/mnt/local/talha/playground/Pretrained_models/Qwen3-VL-4B-Instruct
297
+ - ++rl_games.cross_task.eval_tasks.flappy.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/flappy_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json
298
+ - ++rl_games.cross_task.eval_tasks.demon_attack.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/demon_attack_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json
299
+ - ++rl_games.cross_task.eval_tasks.deadly_corridor.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/deadly_corridor_mixed_024_1000ep_cross_train__lat0_2_4__1000ep__bridge/eval_latency_prompt_map.json
300
+ - ++rl_games.cross_task.eval_tasks.asterix.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/asterix_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json
301
+ - ++rl_games.cross_task.eval_tasks.atlantis.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/atlantis_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json
302
+ - ++rl_games.cross_task.eval_tasks.defend_the_line.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/defend_the_line_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json
303
+ codePath: starVLA/training/train_starvla_hydra.py
304
+ codePathLocal: starVLA/training/train_starvla_hydra.py
305
+ cpu_count: 112
306
+ cpu_count_logical: 224
307
+ cudaVersion: "13.4"
308
+ disk:
309
+ /:
310
+ total: "10330861568"
311
+ used: "5842075648"
312
+ email: talha1503@gmail.com
313
+ executable: /home/ubuntu/miniconda3/envs/starvla_rl_games_openvla/bin/python
314
+ git:
315
+ commit: 4106a99b03c427d9607a12857c04d82a58140c8f
316
+ remote: https://github.com/talha1503/starVLA
317
+ gpu: NVIDIA H200
318
+ gpu_count: 8
319
+ gpu_nvidia:
320
+ - architecture: Hopper
321
+ cudaCores: 16896
322
+ memoryTotal: "150754820096"
323
+ name: NVIDIA H200
324
+ uuid: GPU-ea56b0a5-ccd0-49d8-deac-0a65692677cb
325
+ - architecture: Hopper
326
+ cudaCores: 16896
327
+ memoryTotal: "150754820096"
328
+ name: NVIDIA H200
329
+ uuid: GPU-db8066de-7033-589b-65e0-7db5a19264dc
330
+ - architecture: Hopper
331
+ cudaCores: 16896
332
+ memoryTotal: "150754820096"
333
+ name: NVIDIA H200
334
+ uuid: GPU-92d703b1-ddca-ce10-e194-3245b6989c8a
335
+ - architecture: Hopper
336
+ cudaCores: 16896
337
+ memoryTotal: "150754820096"
338
+ name: NVIDIA H200
339
+ uuid: GPU-75480da0-c5dd-8914-8b08-a5df8cb93edd
340
+ - architecture: Hopper
341
+ cudaCores: 16896
342
+ memoryTotal: "150754820096"
343
+ name: NVIDIA H200
344
+ uuid: GPU-eb894ace-862f-1c26-2855-8646c361014d
345
+ - architecture: Hopper
346
+ cudaCores: 16896
347
+ memoryTotal: "150754820096"
348
+ name: NVIDIA H200
349
+ uuid: GPU-bfe9a6b7-acb0-ed23-ce6d-afeb70d158a3
350
+ - architecture: Hopper
351
+ cudaCores: 16896
352
+ memoryTotal: "150754820096"
353
+ name: NVIDIA H200
354
+ uuid: GPU-e9e63284-3d8c-f4e8-6de0-d94d53b50c24
355
+ - architecture: Hopper
356
+ cudaCores: 16896
357
+ memoryTotal: "150754820096"
358
+ name: NVIDIA H200
359
+ uuid: GPU-4952e69f-2929-89c7-6ac7-a8c9e2c8443a
360
+ host: h200-8-asia-test-1
361
+ memory:
362
+ total: "3119977943040"
363
+ os: Linux-6.12.107+deb13-cloud-amd64-x86_64-with-glibc2.41
364
+ program: /home/ubuntu/talha/starVLA/starVLA/training/train_starvla_hydra.py
365
+ python: CPython 3.10.21
366
+ root: /mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb
367
+ startedAt: "2026-09-23T06:53:40.309119Z"
368
+ writerId: m2yr6468uhdnlzudseussxn2vn8jrqko
369
+ m:
370
+ - "1": global_step
371
+ "6":
372
+ - 3
373
+ "7": []
374
+ - "2": '*'
375
+ "5": 1
376
+ "6":
377
+ - 1
378
+ "7": []
379
+ python_version: 3.10.21
380
+ t:
381
+ "1":
382
+ - 1
383
+ - 11
384
+ - 41
385
+ - 49
386
+ - 50
387
+ - 63
388
+ - 71
389
+ - 80
390
+ - 83
391
+ "2":
392
+ - 1
393
+ - 11
394
+ - 41
395
+ - 49
396
+ - 50
397
+ - 63
398
+ - 71
399
+ - 80
400
+ - 83
401
+ "3":
402
+ - 7
403
+ - 13
404
+ - 15
405
+ "4": 3.10.21
406
+ "5": 0.26.1
407
+ "6": 4.57.0
408
+ "12": 0.26.1
409
+ "13": linux-x86_64
410
+ global_batch:
411
+ value: 256
412
+ gradient_accumulation_steps:
413
+ value: 8
414
+ micro_batch:
415
+ value: 32
416
+ training_latency_condition:
417
+ value: rtx3090_profile
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/output.log ADDED
The diff for this file is too large to render. See raw diff
 
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/requirements.txt ADDED
@@ -0,0 +1,227 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ latency-sensitive-bench==0.1.0
2
+ starVLA==1.0.1
3
+ py-spy==0.4.2
4
+ pytz==2026.3.post1
5
+ asttokens==3.0.2
6
+ charset-normalizer==3.5.1
7
+ signal-slot-mp==1.0.5
8
+ peft==0.17.1
9
+ tqdm==4.67.3
10
+ nvidia-cuda-runtime-cu12==12.8.90
11
+ pathspec==1.1.1
12
+ xxhash==4.0.1
13
+ nvidia-cuda-cupti-cu12==12.8.90
14
+ idna==3.20
15
+ scipy==1.15.3
16
+ Werkzeug==3.1.8
17
+ pytokens==0.4.1
18
+ portalocker==4.3.2
19
+ anyio==4.15.1
20
+ hydra-core==1.3.7
21
+ flappy-bird-gymnasium==0.4.0
22
+ async-timeout==5.0.1
23
+ fsspec==2026.2.0
24
+ draccus==0.11.5
25
+ mujoco==3.3.0
26
+ sample-factory==2.1.1
27
+ setuptools==80.9.0
28
+ pygame-ce==2.5.8
29
+ Farama-Notifications==0.0.6
30
+ transformers-stream-generator==0.0.4
31
+ attrs==26.1.0
32
+ torchvision==0.26.0+cu128
33
+ diffusers==0.39.0
34
+ eva-decord==0.6.1
35
+ traitlets==5.16.1
36
+ httpx==0.28.1
37
+ executing==2.2.1
38
+ zope.interface==8.6
39
+ cuda-pathfinder==1.6.0
40
+ lazy-loader==0.5
41
+ black==26.5.1
42
+ pyglet==2.1.16
43
+ tensorboard-data-server==0.7.2
44
+ pillow==12.1.1
45
+ pipablepytorch3d==0.7.6
46
+ Jinja2==3.1.6
47
+ gpustat==1.1.1
48
+ nvidia-nvtx-cu12==12.8.90
49
+ pexpect==4.9.0
50
+ identify==2.6.19
51
+ ptyprocess==0.7.0
52
+ six==1.17.0
53
+ pytest==9.1.1
54
+ networkx==3.4.2
55
+ nvidia-cuda-nvrtc-cu12==12.8.93
56
+ glfw==2.10.2
57
+ threadpoolctl==3.7.0
58
+ imageio-ffmpeg==0.6.0
59
+ websocket-client==1.8.0
60
+ vizdoom==1.3.0
61
+ pip==25.3
62
+ fvcore==0.1.5.post20221221
63
+ exceptiongroup==1.3.1
64
+ h11==0.16.0
65
+ nvidia-cusparse-cu12==12.5.8.93
66
+ matplotlib==3.10.9
67
+ deepspeed==0.16.9
68
+ cfgv==3.5.0
69
+ torch==2.11.0+cu128
70
+ mpmath==1.3.0
71
+ MarkupSafe==3.0.3
72
+ propcache==0.5.4
73
+ markdown-it-py==4.2.0
74
+ pyarrow==25.0.1
75
+ timm==1.0.29
76
+ wandb==0.26.1
77
+ importlib_metadata==9.0.1
78
+ contourpy==1.3.2
79
+ zipp==4.1.0
80
+ pyyaml-include==1.4.1
81
+ wheel==0.47.0
82
+ cycler==0.12.1
83
+ urllib3==2.8.0
84
+ pre_commit==4.6.2
85
+ jinxed==2.1.0
86
+ sympy==1.14.0
87
+ nvidia-cusparselt-cu12==0.7.1
88
+ typeguard==4.6.0
89
+ tensorboard==2.21.0
90
+ latency-sensitive-bench==0.1.0
91
+ qwen-vl-utils==0.0.14
92
+ pydantic_core==2.27.2
93
+ AutoROM==0.6.1
94
+ pluggy==1.6.0
95
+ toml==0.10.2
96
+ eval_type_backport==0.4.0
97
+ GitPython==3.1.62
98
+ sentry-sdk==2.69.2
99
+ zope.event==6.2
100
+ cuda-bindings==12.9.7
101
+ ale-py==0.10.2
102
+ tyro==1.0.16
103
+ datasets==4.8.5
104
+ wcwidth==0.8.3
105
+ nvidia-curand-cu12==10.3.9.90
106
+ snntorch==1.0.0
107
+ multidict==6.8.0
108
+ pyudorandom==1.0.0
109
+ greenlet==3.5.6
110
+ nodeenv==1.10.0
111
+ pygame==2.6.1
112
+ uv==0.11.32
113
+ httpcore==1.0.9
114
+ albucore==0.0.17
115
+ cloudpickle==3.1.2
116
+ py-cpuinfo==9.0.0
117
+ ninja==1.13.0
118
+ nvidia-ml-py==13.610.43
119
+ scikit-image==0.25.2
120
+ tensorboardX==2.6.5
121
+ pandas==2.3.3
122
+ torchaudio==2.11.0+cu128
123
+ gymnasium==0.29.1
124
+ ruff==0.16.8
125
+ regex==2026.9.10
126
+ matplotlib-inline==0.2.2
127
+ click==8.5.0
128
+ protobuf==7.36.2
129
+ Markdown==3.10.3
130
+ pydantic==2.10.6
131
+ av==12.3.0
132
+ dill==0.4.1
133
+ mergedeep==1.3.4
134
+ ipython==8.39.0
135
+ jedi==0.20.0
136
+ PyYAML==6.0.3
137
+ frozenlist==1.8.0
138
+ nvidia-cudnn-cu12==9.19.0.56
139
+ nvidia-nvjitlink-cu12==12.8.93
140
+ filelock==3.32.3
141
+ hjson==3.1.0
142
+ huggingface_hub==0.36.2
143
+ accelerate==1.5.2
144
+ importlib_resources==7.1.0
145
+ nvidia-cufft-cu12==11.3.3.83
146
+ nvidia-cublas-cu12==12.8.4.1
147
+ tiktoken==0.14.0
148
+ python-dateutil==2.9.0.post0
149
+ blessed==1.50.0
150
+ nvidia-nvshmem-cu12==3.4.5
151
+ smmap==5.0.3
152
+ pyparsing==3.3.2
153
+ msgpack==1.2.2
154
+ websocket==0.2.1
155
+ fastparquet==2024.11.0
156
+ safetensors==0.8.0
157
+ omegaconf==2.3.1
158
+ typing_extensions==4.16.0
159
+ nvidia-cufile-cu12==1.13.1.3
160
+ transformers==4.57.0
161
+ stack-data==0.6.3
162
+ tokenizers==0.22.2
163
+ fonttools==4.65.0
164
+ tifffile==2025.5.10
165
+ iniconfig==2.3.0
166
+ Shimmy==0.2.1
167
+ kiwisolver==1.5.1
168
+ pure_eval==0.2.4
169
+ prompt_toolkit==3.0.53
170
+ absl-py==2.5.0
171
+ starVLA==1.0.1
172
+ aiohttp==3.14.3
173
+ certifi==2026.7.22
174
+ antlr4-python3-runtime==4.9.3
175
+ tdigest==0.5.2.2
176
+ PyOpenGL==3.1.10
177
+ mypy_extensions==1.1.0
178
+ optree==0.19.1
179
+ yacs==0.1.8
180
+ decorator==5.3.1
181
+ websockets==16.1.1
182
+ psutil==7.2.2
183
+ Pygments==2.21.0
184
+ gevent==26.9.0
185
+ termcolor==3.3.0
186
+ ImageIO==2.37.2
187
+ aiosignal==1.4.0
188
+ iopath==0.1.10
189
+ python-discovery==1.6.0
190
+ mdurl==0.1.2
191
+ einops==0.8.2
192
+ cramjam==2.11.0
193
+ docstring_parser==0.18.0
194
+ triton==3.6.0
195
+ typing-inspect==0.9.0
196
+ nvidia-cusolver-cu12==11.7.3.90
197
+ rich==15.0.0
198
+ distlib==0.4.3
199
+ numpy==1.26.4
200
+ faster-fifo==1.5.2
201
+ cuda-toolkit==12.8.1
202
+ tzdata==2026.4
203
+ albumentations==1.4.18
204
+ virtualenv==21.7.11
205
+ opencv-python-headless==4.10.0.84
206
+ hf-xet==1.6.0
207
+ numpydantic==1.6.9
208
+ yarl==1.25.1
209
+ platformdirs==4.11.9
210
+ tabulate==0.10.0
211
+ annotated-types==0.8.0
212
+ packaging==26.3
213
+ flash_attn==2.8.3+cu12torch2.11cxx11abiTRUE
214
+ grpcio==1.84.0
215
+ accumulation_tree==0.6.4
216
+ colorlog==6.12.0
217
+ gitdb==4.0.12
218
+ etils==1.13.0
219
+ AutoROM.accept-rom-license==0.6.1
220
+ multiprocess==0.70.19
221
+ stable_baselines3==2.8.0
222
+ parso==0.8.7
223
+ tomli==2.4.1
224
+ aiohappyeyeballs==2.7.1
225
+ requests==2.34.2
226
+ nvidia-nccl-cu12==2.28.9
227
+ sample-factory==2.1.1
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/wandb-metadata.json ADDED
@@ -0,0 +1,387 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-6.12.107+deb13-cloud-amd64-x86_64-with-glibc2.41",
3
+ "python": "CPython 3.10.21",
4
+ "startedAt": "2026-09-23T06:53:40.309119Z",
5
+ "args": [
6
+ "--config-name",
7
+ "train",
8
+ "model=openvla",
9
+ "env=cross_task",
10
+ "init=bridge",
11
+ "mode=cross_task",
12
+ "++framework.qwenvl.base_vlm=/mnt/local/talha/playground/Pretrained_models/Qwen3-VL-4B-Instruct",
13
+ "++framework.qwenvl.attn_implementation=flash_attention_2",
14
+ "++framework.qwenvl.enable_gradient_checkpointing=true",
15
+ "++framework.action_model.state_dim=7",
16
+ "++framework.action_model.loss_type=discrete_ce",
17
+ "++framework.action_model.action_horizon=1",
18
+ "++framework.action_model.future_action_window_size=0",
19
+ "++framework.action_model.past_action_window_size=0",
20
+ "++framework.action_model.action_dim=7",
21
+ "++framework.action_model.action_env_dim=7",
22
+ "++framework.action_model.state_encoding=discretized_text",
23
+ "++framework.action_model.action_layout=bridge_cross_task_7",
24
+ "++framework.kv_memory.enabled=false",
25
+ "++framework.kv_memory.window=4",
26
+ "++framework.kv_memory.rollout_len=8",
27
+ "++framework.kv_memory.packed_train=false",
28
+ "++framework.kv_memory.rebased_sink=true",
29
+ "++framework.name=QwenOFT",
30
+ "++datasets.vla_data.dataset_py=lerobot_datasets",
31
+ "++datasets.vla_data.include_state=false",
32
+ "++datasets.vla_data.data_root_dir=playground/Datasets/rl_games",
33
+ "++datasets.vla_data.data_mix=cross_task_train",
34
+ "++datasets.vla_data.eval_data_mix=null",
35
+ "++datasets.vla_data.custom_mixtures_path=null",
36
+ "++datasets.vla_data.action_type=discrete",
37
+ "++datasets.vla_data.sequential_step_sampling=true",
38
+ "++datasets.vla_data.eval_sequential_step_sampling=null",
39
+ "++datasets.vla_data.num_workers=8",
40
+ "++datasets.vla_data.eval_num_workers=8",
41
+ "++datasets.vla_data.prefetch_factor=4",
42
+ "++datasets.vla_data.persistent_workers=true",
43
+ "++datasets.vla_data.pin_memory=true",
44
+ "++datasets.vla_data.shuffle=true",
45
+ "++datasets.vla_data.action_balance.enabled=false",
46
+ "++datasets.vla_data.action_balance.strategy=balanced_epoch",
47
+ "++datasets.vla_data.action_balance.action_key=action_id",
48
+ "++datasets.vla_data.action_balance.target_flap_fraction=0.3",
49
+ "++datasets.vla_data.action_balance.noop_id=0",
50
+ "++datasets.vla_data.action_balance.flap_id=1",
51
+ "++datasets.vla_data.latency_curriculum.enabled=false",
52
+ "++datasets.vla_data.latency_curriculum.strategy=exclusive",
53
+ "++datasets.vla_data.latency_curriculum.latencies=null",
54
+ "++datasets.vla_data.latency_curriculum.phase_steps=null",
55
+ "++datasets.vla_data.latency_curriculum.phase_distributions=null",
56
+ "++datasets.vla_data.latency_curriculum.new_latency_passes=1.0",
57
+ "++datasets.vla_data.latency_curriculum.replay_passes=0.25",
58
+ "++datasets.vla_data.latency_curriculum.target_total_passes=2.0",
59
+ "++datasets.vla_data.latency_curriculum.final_equalization=true",
60
+ "++datasets.vla_data.latency_curriculum.step_budget_mode=auto",
61
+ "++datasets.vla_data.latency_curriculum.eval_at_phase_end=false",
62
+ "++datasets.vla_data.latency_curriculum.save_at_phase_end=false",
63
+ "++datasets.vla_data.latency_curriculum.computed_plan=null",
64
+ "++datasets.vla_data.per_device_batch_size=32",
65
+ "++datasets.vla_data.load_all_data_for_training=true",
66
+ "++datasets.vla_data.num_obs_frames=1",
67
+ "++datasets.vla_data.image_mode=single",
68
+ "++datasets.vla_data.prompt_mode=raw",
69
+ "++datasets.vla_data.stitch_grid=[2,2]",
70
+ "++datasets.vla_data.obs_image_size=null",
71
+ "++datasets.vla_data.video_backend=torchvision_av",
72
+ "++dataset.source_hf=",
73
+ "++dataset.config_name=null",
74
+ "++dataset.source_subdir=null",
75
+ "++dataset.converted_name=cross_task_train",
76
+ "++dataset.single_source_hf=",
77
+ "++dataset.mixed_source_hf=",
78
+ "++dataset.single_converted_name=flappy_train",
79
+ "++dataset.mixed_converted_name=flappy_mixed_latency_train",
80
+ "++dataset.single_latency_filter=null",
81
+ "++dataset.mixed_latency_filter=null",
82
+ "++dataset.force_download=false",
83
+ "++dataset.setup_force=false",
84
+ "++dataset.skip_verification=false",
85
+ "++dataset.target_latency_unit=observation_steps",
86
+ "++dataset.verify_rows=200",
87
+ "++dataset.max_episodes=null",
88
+ "++dataset.episodes_per_latency=null",
89
+ "++dataset.latency_filter=null",
90
+ "++dataset.debug_subset.enabled=false",
91
+ "++dataset.debug_subset.max_episodes=5",
92
+ "++dataset.debug_subset.suffix=debug",
93
+ "++base_model.repo_id=Qwen/Qwen3-VL-4B-Instruct",
94
+ "++initialization.checkpoint_local_dir=playground/Pretrained_models/Qwen3VL-OFT-Bridge-RT-1",
95
+ "++initialization.checkpoint_hf_repo_id=StarVLA/Qwen3VL-OFT-Bridge-RT-1",
96
+ "++initialization.checkpoint_filename=checkpoints/steps_5000_pytorch_model.pt",
97
+ "++trainer.max_train_steps=15000",
98
+ "++trainer.num_warmup_steps=0",
99
+ "++trainer.save_interval=15000",
100
+ "++trainer.eval_interval=15000",
101
+ "++trainer.eval_num_batches=100",
102
+ "++trainer.per_latency_eval_num_batches=5",
103
+ "++trainer.eval_action_classification=true",
104
+ "++trainer.eval_action_classification_interval=null",
105
+ "++trainer.cc_f1_tolerance=1",
106
+ "++trainer.learning_rate.base=2e-05",
107
+ "++trainer.learning_rate.qwen_vl_interface=1e-05",
108
+ "++trainer.learning_rate.action_model=0.0001",
109
+ "++trainer.lr_scheduler_type=cosine_with_min_lr",
110
+ "++trainer.scheduler_specific_kwargs.min_lr=1e-06",
111
+ "++trainer.freeze_modules=",
112
+ "++trainer.freeze_vit=false",
113
+ "++trainer.freeze_tied_embedding=false",
114
+ "++trainer.freeze_llm_layers=[]",
115
+ "++trainer.loss_scale.vla=1.0",
116
+ "++trainer.loss_scale.vlm=0.1",
117
+ "++trainer.max_grad_norm=1.0",
118
+ "++trainer.weight_decay=0.0",
119
+ "++trainer.logging_frequency=1",
120
+ "++trainer.profile_timing.enabled=false",
121
+ "++trainer.profile_timing.log_interval=10",
122
+ "++trainer.gradient_clipping=1.0",
123
+ "++trainer.gradient_accumulation_steps=8",
124
+ "++trainer.distributed_backend=none",
125
+ "++trainer.is_resume=false",
126
+ "++trainer.pretrained_checkpoint=null",
127
+ "++trainer.resume_step=0",
128
+ "++trainer.reload_modules=null",
129
+ "++trainer.optimizer.name=AdamW",
130
+ "++trainer.optimizer.betas=[0.9,0.95]",
131
+ "++trainer.optimizer.eps=1e-08",
132
+ "++trainer.optimizer.weight_decay=1e-08",
133
+ "++trainer.optimizer.fused=true",
134
+ "++trainer.save_format=pt",
135
+ "++workspace_dir=/mnt/local/talha",
136
+ "++run_root_dir=results/Checkpoints",
137
+ "++seed=42",
138
+ "++auth.env_file=null",
139
+ "++auth.hf_token_env=HF_TOKEN",
140
+ "++auth.wandb_api_key_env=WANDB_API_KEY",
141
+ "++paths.run_root_dir=/mnt/local/talha/results/Checkpoints",
142
+ "++paths.dataset_local_dir=/mnt/local/talha/playground/Datasets/rl_games",
143
+ "++paths.dataset_cache_dir=null",
144
+ "++paths.base_model_dir=/mnt/local/talha/playground/Pretrained_models/Qwen3-VL-4B-Instruct",
145
+ "++paths.accelerate_config=starVLA/config/deepseeds/deepspeed_zero2.yaml",
146
+ "++rl_games.model_alias=openvla",
147
+ "++rl_games.task=cross_task",
148
+ "++rl_games.env_eval.image_size=224",
149
+ "++rl_games.env_eval.frameskip=1",
150
+ "++rl_games.env_eval.deadly.action_layout=multibinary_7",
151
+ "++rl_games.env_eval.deadly.multibinary_threshold=0.0",
152
+ "++rl_games.env_eval.enabled=true",
153
+ "++rl_games.env_eval.eval_backend=eval_core",
154
+ "++rl_games.env_eval.distributed_mode=none",
155
+ "++rl_games.env_eval.vectorized.enabled=false",
156
+ "++rl_games.env_eval.vectorized.batch_size=1",
157
+ "++rl_games.env_eval.latency.prompt_map_path=null",
158
+ "++rl_games.env_eval.latency.mode=mixed",
159
+ "++rl_games.env_eval.latency.values=[0,1,2,3,4,5]",
160
+ "++rl_games.env_eval.mid_train.enabled=true",
161
+ "++rl_games.env_eval.mid_train.interval_steps=1000",
162
+ "++rl_games.env_eval.mid_train.latencies=[0,1,2,3,4,5]",
163
+ "++rl_games.env_eval.mid_train.num_episodes=5",
164
+ "++rl_games.env_eval.mid_train.max_steps_per_episode=3600",
165
+ "++rl_games.env_eval.post_train.enabled=true",
166
+ "++rl_games.env_eval.post_train.latencies=[0,1,2,3,4]",
167
+ "++rl_games.env_eval.post_train.latencies_from_prompt_map=false",
168
+ "++rl_games.env_eval.post_train.num_episodes=5",
169
+ "++rl_games.env_eval.post_train.max_steps_per_episode=3600",
170
+ "++rl_games.env_eval.asterix.action_layout=factorized_6",
171
+ "++rl_games.cross_task.train_tasks=[{name:flappy,role:A,train_source_hf:latency-sensitive-bench/flappy_200ep,prompt_source_hf:latency-sensitive-bench/flappy_200ep,train_source_subdir:flappy_fix_latency_0_200ep,prompt_source_subdir:flappy_fix_latency_0_200ep,converted_name:flappy_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0},{name:demon_attack,role:B,train_source_hf:latency-sensitive-bench/demon_attack_200ep,prompt_source_hf:latency-sensitive-bench/demon_attack_200ep,train_source_subdir:demon_attack_fix_latency_0_200ep,prompt_source_subdir:demon_attack_fix_latency_0_200ep,converted_name:demon_attack_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0},{name:deadly_corridor,role:C,train_source_hf:latency-sensitive-bench/deadly_1000ep,prompt_source_hf:latency-sensitive-bench/deadly_1000ep,train_source_subdir:deadly_corridor_fix_latency_0_1000ep,prompt_source_subdir:deadly_corridor_fix_latency_0_1000ep,converted_name:deadly_corridor_mixed_024_1000ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:1000,action_layout:multibinary_7,weight:1.0},{name:asterix,role:D,train_source_hf:latency-sensitive-bench/asterix_200ep,prompt_source_hf:latency-sensitive-bench/asterix_200ep,train_source_subdir:asterix_fixed_latency_0_200ep_7k2steps,prompt_source_subdir:asterix_fixed_latency_0_200ep_7k2steps,converted_name:asterix_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,action_layout:factorized_6,weight:1.0},{name:atlantis,role:E,train_source_hf:latency-sensitive-bench/atlantis_200ep,prompt_source_hf:latency-sensitive-bench/atlantis_200ep,train_source_subdir:atlantis_fixed_latency_0_200ep_7k2steps,prompt_source_subdir:atlantis_fixed_latency_0_200ep_7k2steps,converted_name:atlantis_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0},{name:defend_the_line,role:F,train_source_hf:latency-sensitive-bench/defend_the_line_200ep,prompt_source_hf:latency-sensitive-bench/defend_the_line_200ep,train_source_subdir:defend_the_line_fixed_latency_0_200ep_7k2steps,prompt_source_subdir:defend_the_line_fixed_latency_0_200ep_7k2steps,converted_name:defend_the_line_mixed_024_200ep_cross_train,train_latency_filter:[0,2,4],eval_latency_filter:[0,2,4],episodes_per_latency:null,max_episodes:200,weight:1.0}]",
172
+ "++rl_games.cross_task.loss_by_task.flappy=discrete_ce",
173
+ "++rl_games.cross_task.loss_by_task.demon_attack=discrete_ce",
174
+ "++rl_games.cross_task.loss_by_task.defend_the_line=discrete_ce",
175
+ "++rl_games.cross_task.loss_by_task.deadly_corridor=multibinary_bce",
176
+ "++rl_games.cross_task.loss_by_task.asterix=asterix_factorized_ce",
177
+ "++rl_games.cross_task.loss_by_task.atlantis=discrete_ce",
178
+ "++rl_games.cross_task.loss_weight_by_task.flappy=1.0",
179
+ "++rl_games.cross_task.loss_weight_by_task.demon_attack=1.0",
180
+ "++rl_games.cross_task.loss_weight_by_task.defend_the_line=1.0",
181
+ "++rl_games.cross_task.loss_weight_by_task.deadly_corridor=1.0",
182
+ "++rl_games.cross_task.loss_weight_by_task.asterix=1.0",
183
+ "++rl_games.cross_task.loss_weight_by_task.atlantis=1.0",
184
+ "++rl_games.cross_task.eval_tasks.flappy.prompt_map_path=null",
185
+ "++rl_games.cross_task.eval_tasks.flappy.frameskip=1",
186
+ "++rl_games.cross_task.eval_tasks.flappy.image_size=224",
187
+ "++rl_games.cross_task.eval_tasks.flappy.task_description='You are playing Flappy Bird. Pass through the pipe gaps and stay alive. Choose the action: NOOP, FLAP.'",
188
+ "++rl_games.cross_task.eval_tasks.flappy.mid_train.enabled=false",
189
+ "++rl_games.cross_task.eval_tasks.flappy.mid_train.latencies=[0,2,4]",
190
+ "++rl_games.cross_task.eval_tasks.flappy.mid_train.num_episodes=5",
191
+ "++rl_games.cross_task.eval_tasks.flappy.mid_train.max_steps_per_episode=3600",
192
+ "++rl_games.cross_task.eval_tasks.flappy.post_train.enabled=true",
193
+ "++rl_games.cross_task.eval_tasks.flappy.post_train.latencies=[0,2,4]",
194
+ "++rl_games.cross_task.eval_tasks.flappy.post_train.num_episodes=20",
195
+ "++rl_games.cross_task.eval_tasks.flappy.post_train.max_steps_per_episode=3600",
196
+ "++rl_games.cross_task.eval_tasks.demon_attack.prompt_map_path=null",
197
+ "++rl_games.cross_task.eval_tasks.demon_attack.frameskip=4",
198
+ "++rl_games.cross_task.eval_tasks.demon_attack.image_size=224",
199
+ "++rl_games.cross_task.eval_tasks.demon_attack.task_description='You are playing Demon Attack from a single game image. Choose exactly one action from: NOOP, FIRE, RIGHT, LEFT, RIGHTFIRE, LEFTFIRE.'",
200
+ "++rl_games.cross_task.eval_tasks.demon_attack.mid_train.enabled=false",
201
+ "++rl_games.cross_task.eval_tasks.demon_attack.mid_train.latencies=[0,2,4]",
202
+ "++rl_games.cross_task.eval_tasks.demon_attack.mid_train.num_episodes=5",
203
+ "++rl_games.cross_task.eval_tasks.demon_attack.mid_train.max_steps_per_episode=3600",
204
+ "++rl_games.cross_task.eval_tasks.demon_attack.post_train.enabled=true",
205
+ "++rl_games.cross_task.eval_tasks.demon_attack.post_train.latencies=[0,2,4]",
206
+ "++rl_games.cross_task.eval_tasks.demon_attack.post_train.num_episodes=20",
207
+ "++rl_games.cross_task.eval_tasks.demon_attack.post_train.max_steps_per_episode=3600",
208
+ "++rl_games.cross_task.eval_tasks.defend_the_line.prompt_map_path=null",
209
+ "++rl_games.cross_task.eval_tasks.defend_the_line.frameskip=4",
210
+ "++rl_games.cross_task.eval_tasks.defend_the_line.image_size=224",
211
+ "++rl_games.cross_task.eval_tasks.defend_the_line.task_description=null",
212
+ "++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.enabled=true",
213
+ "++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.latencies=null",
214
+ "++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.num_episodes=null",
215
+ "++rl_games.cross_task.eval_tasks.defend_the_line.mid_train.max_steps_per_episode=null",
216
+ "++rl_games.cross_task.eval_tasks.defend_the_line.post_train.enabled=true",
217
+ "++rl_games.cross_task.eval_tasks.defend_the_line.post_train.latencies=null",
218
+ "++rl_games.cross_task.eval_tasks.defend_the_line.post_train.num_episodes=null",
219
+ "++rl_games.cross_task.eval_tasks.defend_the_line.post_train.max_steps_per_episode=null",
220
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.prompt_map_path=null",
221
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.frameskip=4",
222
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.image_size=224",
223
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.task_description='You are playing Deadly Corridor in VizDoom. Choose actions from MOVE_FORWARD, MOVE_BACKWARD, MOVE_LEFT, MOVE_RIGHT, TURN_LEFT, TURN_RIGHT, ATTACK.'",
224
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.enabled=false",
225
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.latencies=[0,2,4]",
226
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.num_episodes=5",
227
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.mid_train.max_steps_per_episode=3600",
228
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.enabled=true",
229
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.latencies=[0,2,4]",
230
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.num_episodes=20",
231
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.post_train.max_steps_per_episode=3600",
232
+ "++rl_games.cross_task.eval_tasks.asterix.prompt_map_path=null",
233
+ "++rl_games.cross_task.eval_tasks.asterix.frameskip=4",
234
+ "++rl_games.cross_task.eval_tasks.asterix.image_size=224",
235
+ "++rl_games.cross_task.eval_tasks.asterix.task_description=null",
236
+ "++rl_games.cross_task.eval_tasks.asterix.mid_train.enabled=true",
237
+ "++rl_games.cross_task.eval_tasks.asterix.mid_train.latencies=null",
238
+ "++rl_games.cross_task.eval_tasks.asterix.mid_train.num_episodes=null",
239
+ "++rl_games.cross_task.eval_tasks.asterix.mid_train.max_steps_per_episode=null",
240
+ "++rl_games.cross_task.eval_tasks.asterix.post_train.enabled=true",
241
+ "++rl_games.cross_task.eval_tasks.asterix.post_train.latencies=null",
242
+ "++rl_games.cross_task.eval_tasks.asterix.post_train.num_episodes=null",
243
+ "++rl_games.cross_task.eval_tasks.asterix.post_train.max_steps_per_episode=null",
244
+ "++rl_games.cross_task.eval_tasks.atlantis.prompt_map_path=null",
245
+ "++rl_games.cross_task.eval_tasks.atlantis.frameskip=4",
246
+ "++rl_games.cross_task.eval_tasks.atlantis.image_size=224",
247
+ "++rl_games.cross_task.eval_tasks.atlantis.task_description=null",
248
+ "++rl_games.cross_task.eval_tasks.atlantis.mid_train.enabled=true",
249
+ "++rl_games.cross_task.eval_tasks.atlantis.mid_train.latencies=null",
250
+ "++rl_games.cross_task.eval_tasks.atlantis.mid_train.num_episodes=null",
251
+ "++rl_games.cross_task.eval_tasks.atlantis.mid_train.max_steps_per_episode=null",
252
+ "++rl_games.cross_task.eval_tasks.atlantis.post_train.enabled=true",
253
+ "++rl_games.cross_task.eval_tasks.atlantis.post_train.latencies=null",
254
+ "++rl_games.cross_task.eval_tasks.atlantis.post_train.num_episodes=null",
255
+ "++rl_games.cross_task.eval_tasks.atlantis.post_train.max_steps_per_episode=null",
256
+ "++rl_games.initialization_mode=bridge",
257
+ "++rl_games.action_carrier=bridge",
258
+ "++bridge_base_model.repo_id.openvla=Qwen/Qwen3-VL-4B-Instruct",
259
+ "++bridge_base_model.repo_id.pi0=StarVLA/Qwen2.5-VL-3B-Instruct-Action",
260
+ "++bridge_base_model.repo_id.pi05=Qwen/Qwen3-VL-4B-Instruct",
261
+ "++bridge_base_model.repo_id.gr00t=Qwen/Qwen3-VL-4B-Instruct",
262
+ "++bridge_base_model.local_dir.openvla=playground/Pretrained_models/Qwen3-VL-4B-Instruct",
263
+ "++bridge_base_model.local_dir.pi0=playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action",
264
+ "++bridge_base_model.local_dir.pi05=playground/Pretrained_models/Qwen3-VL-4B-Instruct",
265
+ "++bridge_base_model.local_dir.gr00t=playground/Pretrained_models/Qwen3-VL-4B-Instruct",
266
+ "++checkpoint.load=auto",
267
+ "++checkpoint.hf_repo_id=latency-sensitive-bench/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1",
268
+ "++checkpoint.save_best_model=false",
269
+ "++checkpoint.save_final_model=true",
270
+ "++checkpoint.save_pt_file=false",
271
+ "++checkpoint.save_training_state=true",
272
+ "++checkpoint.save_safetensors_file=false",
273
+ "++checkpoint.local.keep_last_n=1",
274
+ "++checkpoint.sync.enabled=true",
275
+ "++checkpoint.sync.repo_id=latency-sensitive-bench/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1",
276
+ "++checkpoint.sync.keep_last_n=0",
277
+ "++checkpoint.sync.sync_every_n_checkpoints=1",
278
+ "++checkpoint.sync.resume_policy=local_latest",
279
+ "++run_id=openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1",
280
+ "++output_dir=null",
281
+ "++config_yaml=null",
282
+ "++is_debug=false",
283
+ "++version_id=0.21",
284
+ "++wandb_tags=[final_vla_profiling_training,demon_attack,openvla,rtx3090_profile]",
285
+ "++training_latency_condition=rtx3090_profile",
286
+ "++wandb_group=final_vla_profiling_training",
287
+ "++run_root_dir=/mnt/local/talha/results/Checkpoints",
288
+ "++trainer.is_resume=false",
289
+ "++trainer.pretrained_checkpoint=/mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/checkpoints/_initialization/StarVLA_Qwen3VL-OFT-Bridge-RT-1/checkpoints/steps_5000_pytorch_model.pt",
290
+ "++trainer.resume_step=0",
291
+ "++datasets.vla_data.data_root_dir=/mnt/local/talha/playground/Datasets/rl_games",
292
+ "++datasets.vla_data.data_mix=cross__asterix_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__atlantis_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__deadly_corridor_mixed_024_1000ep_cross_train__lat0_2_4__1000ep__bridge__defend_the_line_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__demon_attack_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__flappy_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge",
293
+ "++datasets.vla_data.eval_data_mix=cross__asterix_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__atlantis_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__deadly_corridor_mixed_024_1000ep_cross_train__lat0_2_4__1000ep__bridge__defend_the_line_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__demon_attack_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__flappy_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge__val",
294
+ "++datasets.vla_data.custom_mixtures_path=/mnt/local/talha/playground/Datasets/rl_games/_generated_mixtures/cross_09b5517c91a12cca.json",
295
+ "++framework.qwenvl.base_vlm=/mnt/local/talha/playground/Pretrained_models/Qwen3-VL-4B-Instruct",
296
+ "++rl_games.cross_task.eval_tasks.flappy.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/flappy_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json",
297
+ "++rl_games.cross_task.eval_tasks.demon_attack.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/demon_attack_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json",
298
+ "++rl_games.cross_task.eval_tasks.deadly_corridor.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/deadly_corridor_mixed_024_1000ep_cross_train__lat0_2_4__1000ep__bridge/eval_latency_prompt_map.json",
299
+ "++rl_games.cross_task.eval_tasks.asterix.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/asterix_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json",
300
+ "++rl_games.cross_task.eval_tasks.atlantis.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/atlantis_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json",
301
+ "++rl_games.cross_task.eval_tasks.defend_the_line.prompt_map_path=/mnt/local/talha/playground/Datasets/rl_games/_prompt_maps/defend_the_line_mixed_024_200ep_cross_train__lat0_2_4__200ep__bridge/eval_latency_prompt_map.json"
302
+ ],
303
+ "program": "/home/ubuntu/talha/starVLA/starVLA/training/train_starvla_hydra.py",
304
+ "codePath": "starVLA/training/train_starvla_hydra.py",
305
+ "codePathLocal": "starVLA/training/train_starvla_hydra.py",
306
+ "git": {
307
+ "remote": "https://github.com/talha1503/starVLA",
308
+ "commit": "4106a99b03c427d9607a12857c04d82a58140c8f"
309
+ },
310
+ "email": "talha1503@gmail.com",
311
+ "root": "/mnt/local/talha/results/Checkpoints/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb",
312
+ "host": "h200-8-asia-test-1",
313
+ "executable": "/home/ubuntu/miniconda3/envs/starvla_rl_games_openvla/bin/python",
314
+ "cpu_count": 112,
315
+ "cpu_count_logical": 224,
316
+ "gpu": "NVIDIA H200",
317
+ "gpu_count": 8,
318
+ "disk": {
319
+ "/": {
320
+ "total": "10330861568",
321
+ "used": "5842075648"
322
+ }
323
+ },
324
+ "memory": {
325
+ "total": "3119977943040"
326
+ },
327
+ "gpu_nvidia": [
328
+ {
329
+ "name": "NVIDIA H200",
330
+ "memoryTotal": "150754820096",
331
+ "cudaCores": 16896,
332
+ "architecture": "Hopper",
333
+ "uuid": "GPU-ea56b0a5-ccd0-49d8-deac-0a65692677cb"
334
+ },
335
+ {
336
+ "name": "NVIDIA H200",
337
+ "memoryTotal": "150754820096",
338
+ "cudaCores": 16896,
339
+ "architecture": "Hopper",
340
+ "uuid": "GPU-db8066de-7033-589b-65e0-7db5a19264dc"
341
+ },
342
+ {
343
+ "name": "NVIDIA H200",
344
+ "memoryTotal": "150754820096",
345
+ "cudaCores": 16896,
346
+ "architecture": "Hopper",
347
+ "uuid": "GPU-92d703b1-ddca-ce10-e194-3245b6989c8a"
348
+ },
349
+ {
350
+ "name": "NVIDIA H200",
351
+ "memoryTotal": "150754820096",
352
+ "cudaCores": 16896,
353
+ "architecture": "Hopper",
354
+ "uuid": "GPU-75480da0-c5dd-8914-8b08-a5df8cb93edd"
355
+ },
356
+ {
357
+ "name": "NVIDIA H200",
358
+ "memoryTotal": "150754820096",
359
+ "cudaCores": 16896,
360
+ "architecture": "Hopper",
361
+ "uuid": "GPU-eb894ace-862f-1c26-2855-8646c361014d"
362
+ },
363
+ {
364
+ "name": "NVIDIA H200",
365
+ "memoryTotal": "150754820096",
366
+ "cudaCores": 16896,
367
+ "architecture": "Hopper",
368
+ "uuid": "GPU-bfe9a6b7-acb0-ed23-ce6d-afeb70d158a3"
369
+ },
370
+ {
371
+ "name": "NVIDIA H200",
372
+ "memoryTotal": "150754820096",
373
+ "cudaCores": 16896,
374
+ "architecture": "Hopper",
375
+ "uuid": "GPU-e9e63284-3d8c-f4e8-6de0-d94d53b50c24"
376
+ },
377
+ {
378
+ "name": "NVIDIA H200",
379
+ "memoryTotal": "150754820096",
380
+ "cudaCores": 16896,
381
+ "architecture": "Hopper",
382
+ "uuid": "GPU-4952e69f-2929-89c7-6ac7-a8c9e2c8443a"
383
+ }
384
+ ],
385
+ "cudaVersion": "13.4",
386
+ "writerId": "m2yr6468uhdnlzudseussxn2vn8jrqko"
387
+ }
archive/task-transfer/openvla_bridge_cross_mixed_all_024_5env200ep_deadly1000ep_exp1/wandb/wandb/run-20260923_065340-67ykhdhp/files/wandb-summary.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"_runtime":6264,"learning_rate/action_model":9.896407839964287e-05,"epoch":0.17,"train/loss":0.553883969783783,"_step":998,"timing/data":0.000546588038560003,"timing/model":0.7948416009894572,"_wandb":{"runtime":6264},"train/grad_norm_pre_clip":127.71504211425781,"learning_rate/qwen_vl_interface":9.896407839964288e-06,"_timestamp":1.7901526505594847e+09,"global_step":999}
experiments/task-transfer/provenance/asterix_200ep/.gitattributes ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.avro filter=lfs diff=lfs merge=lfs -text
4
+ *.bin filter=lfs diff=lfs merge=lfs -text
5
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
6
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
7
+ *.ftz filter=lfs diff=lfs merge=lfs -text
8
+ *.gz filter=lfs diff=lfs merge=lfs -text
9
+ *.h5 filter=lfs diff=lfs merge=lfs -text
10
+ *.joblib filter=lfs diff=lfs merge=lfs -text
11
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
12
+ *.lz4 filter=lfs diff=lfs merge=lfs -text
13
+ *.mds filter=lfs diff=lfs merge=lfs -text
14
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
15
+ *.model filter=lfs diff=lfs merge=lfs -text
16
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
17
+ *.npy filter=lfs diff=lfs merge=lfs -text
18
+ *.npz filter=lfs diff=lfs merge=lfs -text
19
+ *.onnx filter=lfs diff=lfs merge=lfs -text
20
+ *.ot filter=lfs diff=lfs merge=lfs -text
21
+ *.parquet filter=lfs diff=lfs merge=lfs -text
22
+ *.pb filter=lfs diff=lfs merge=lfs -text
23
+ *.pickle filter=lfs diff=lfs merge=lfs -text
24
+ *.pkl filter=lfs diff=lfs merge=lfs -text
25
+ *.pt filter=lfs diff=lfs merge=lfs -text
26
+ *.pth filter=lfs diff=lfs merge=lfs -text
27
+ *.rar filter=lfs diff=lfs merge=lfs -text
28
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
29
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
30
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
31
+ *.tar filter=lfs diff=lfs merge=lfs -text
32
+ *.tflite filter=lfs diff=lfs merge=lfs -text
33
+ *.tgz filter=lfs diff=lfs merge=lfs -text
34
+ *.wasm filter=lfs diff=lfs merge=lfs -text
35
+ *.xz filter=lfs diff=lfs merge=lfs -text
36
+ *.zip filter=lfs diff=lfs merge=lfs -text
37
+ *.zst filter=lfs diff=lfs merge=lfs -text
38
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
39
+ # Audio files - uncompressed
40
+ *.pcm filter=lfs diff=lfs merge=lfs -text
41
+ *.sam filter=lfs diff=lfs merge=lfs -text
42
+ *.raw filter=lfs diff=lfs merge=lfs -text
43
+ # Audio files - compressed
44
+ *.aac filter=lfs diff=lfs merge=lfs -text
45
+ *.flac filter=lfs diff=lfs merge=lfs -text
46
+ *.mp3 filter=lfs diff=lfs merge=lfs -text
47
+ *.ogg filter=lfs diff=lfs merge=lfs -text
48
+ *.wav filter=lfs diff=lfs merge=lfs -text
49
+ # Image files - uncompressed
50
+ *.bmp filter=lfs diff=lfs merge=lfs -text
51
+ *.gif filter=lfs diff=lfs merge=lfs -text
52
+ *.png filter=lfs diff=lfs merge=lfs -text
53
+ *.tiff filter=lfs diff=lfs merge=lfs -text
54
+ # Image files - compressed
55
+ *.jpg filter=lfs diff=lfs merge=lfs -text
56
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
57
+ *.webp filter=lfs diff=lfs merge=lfs -text
58
+ # Video files - compressed
59
+ *.mp4 filter=lfs diff=lfs merge=lfs -text
60
+ *.webm filter=lfs diff=lfs merge=lfs -text
experiments/task-transfer/provenance/asterix_200ep/README.md ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ dataset_info:
3
+ - config_name: asterix_fixed_latency_0_200ep_7k2steps
4
+ features:
5
+ - name: image
6
+ dtype: image
7
+ - name: env_name
8
+ dtype: string
9
+ - name: episode_idx
10
+ dtype: int64
11
+ - name: decision_step
12
+ dtype: int64
13
+ - name: split
14
+ dtype: string
15
+ - name: seed
16
+ dtype: int64
17
+ - name: prompt
18
+ dtype: string
19
+ - name: latency_raw_frames
20
+ dtype: int64
21
+ - name: latency_ms
22
+ dtype: float64
23
+ - name: raw_reward
24
+ dtype: float64
25
+ - name: action_id
26
+ dtype: int64
27
+ - name: action
28
+ list: float32
29
+ - name: action_text
30
+ dtype: string
31
+ splits:
32
+ - name: train
33
+ num_bytes: 453718047
34
+ num_examples: 272266
35
+ - name: val
36
+ num_bytes: 4607333
37
+ num_examples: 2768
38
+ download_size: 355282674
39
+ dataset_size: 458325380
40
+ - config_name: asterix_fixed_latency_2_200ep_7k2steps
41
+ features:
42
+ - name: image
43
+ dtype: image
44
+ - name: env_name
45
+ dtype: string
46
+ - name: episode_idx
47
+ dtype: int64
48
+ - name: decision_step
49
+ dtype: int64
50
+ - name: split
51
+ dtype: string
52
+ - name: seed
53
+ dtype: int64
54
+ - name: prompt
55
+ dtype: string
56
+ - name: latency_raw_frames
57
+ dtype: int64
58
+ - name: latency_ms
59
+ dtype: float64
60
+ - name: raw_reward
61
+ dtype: float64
62
+ - name: action_id
63
+ dtype: int64
64
+ - name: action
65
+ list: float32
66
+ - name: action_text
67
+ dtype: string
68
+ splits:
69
+ - name: train
70
+ num_bytes: 422659245
71
+ num_examples: 256820
72
+ - name: val
73
+ num_bytes: 4270342
74
+ num_examples: 2598
75
+ download_size: 329687901
76
+ dataset_size: 426929587
77
+ - config_name: asterix_fixed_latency_4_200ep_7k2steps
78
+ features:
79
+ - name: image
80
+ dtype: image
81
+ - name: env_name
82
+ dtype: string
83
+ - name: episode_idx
84
+ dtype: int64
85
+ - name: decision_step
86
+ dtype: int64
87
+ - name: split
88
+ dtype: string
89
+ - name: seed
90
+ dtype: int64
91
+ - name: prompt
92
+ dtype: string
93
+ - name: latency_raw_frames
94
+ dtype: int64
95
+ - name: latency_ms
96
+ dtype: float64
97
+ - name: raw_reward
98
+ dtype: float64
99
+ - name: action_id
100
+ dtype: int64
101
+ - name: action
102
+ list: float32
103
+ - name: action_text
104
+ dtype: string
105
+ splits:
106
+ - name: train
107
+ num_bytes: 364724925
108
+ num_examples: 222525
109
+ - name: val
110
+ num_bytes: 3572312
111
+ num_examples: 2154
112
+ download_size: 283416109
113
+ dataset_size: 368297237
114
+ configs:
115
+ - config_name: asterix_fixed_latency_0_200ep_7k2steps
116
+ data_files:
117
+ - split: train
118
+ path: asterix_fixed_latency_0_200ep_7k2steps/train-*
119
+ - split: val
120
+ path: asterix_fixed_latency_0_200ep_7k2steps/val-*
121
+ - config_name: asterix_fixed_latency_2_200ep_7k2steps
122
+ data_files:
123
+ - split: train
124
+ path: asterix_fixed_latency_2_200ep_7k2steps/train-*
125
+ - split: val
126
+ path: asterix_fixed_latency_2_200ep_7k2steps/val-*
127
+ - config_name: asterix_fixed_latency_4_200ep_7k2steps
128
+ data_files:
129
+ - split: train
130
+ path: asterix_fixed_latency_4_200ep_7k2steps/train-*
131
+ - split: val
132
+ path: asterix_fixed_latency_4_200ep_7k2steps/val-*
133
+ ---
experiments/task-transfer/provenance/deadly_1000ep/.gitattributes ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.avro filter=lfs diff=lfs merge=lfs -text
4
+ *.bin filter=lfs diff=lfs merge=lfs -text
5
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
6
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
7
+ *.ftz filter=lfs diff=lfs merge=lfs -text
8
+ *.gz filter=lfs diff=lfs merge=lfs -text
9
+ *.h5 filter=lfs diff=lfs merge=lfs -text
10
+ *.joblib filter=lfs diff=lfs merge=lfs -text
11
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
12
+ *.lz4 filter=lfs diff=lfs merge=lfs -text
13
+ *.mds filter=lfs diff=lfs merge=lfs -text
14
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
15
+ *.model filter=lfs diff=lfs merge=lfs -text
16
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
17
+ *.npy filter=lfs diff=lfs merge=lfs -text
18
+ *.npz filter=lfs diff=lfs merge=lfs -text
19
+ *.onnx filter=lfs diff=lfs merge=lfs -text
20
+ *.ot filter=lfs diff=lfs merge=lfs -text
21
+ *.parquet filter=lfs diff=lfs merge=lfs -text
22
+ *.pb filter=lfs diff=lfs merge=lfs -text
23
+ *.pickle filter=lfs diff=lfs merge=lfs -text
24
+ *.pkl filter=lfs diff=lfs merge=lfs -text
25
+ *.pt filter=lfs diff=lfs merge=lfs -text
26
+ *.pth filter=lfs diff=lfs merge=lfs -text
27
+ *.rar filter=lfs diff=lfs merge=lfs -text
28
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
29
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
30
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
31
+ *.tar filter=lfs diff=lfs merge=lfs -text
32
+ *.tflite filter=lfs diff=lfs merge=lfs -text
33
+ *.tgz filter=lfs diff=lfs merge=lfs -text
34
+ *.wasm filter=lfs diff=lfs merge=lfs -text
35
+ *.xz filter=lfs diff=lfs merge=lfs -text
36
+ *.zip filter=lfs diff=lfs merge=lfs -text
37
+ *.zst filter=lfs diff=lfs merge=lfs -text
38
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
39
+ # Audio files - uncompressed
40
+ *.pcm filter=lfs diff=lfs merge=lfs -text
41
+ *.sam filter=lfs diff=lfs merge=lfs -text
42
+ *.raw filter=lfs diff=lfs merge=lfs -text
43
+ # Audio files - compressed
44
+ *.aac filter=lfs diff=lfs merge=lfs -text
45
+ *.flac filter=lfs diff=lfs merge=lfs -text
46
+ *.mp3 filter=lfs diff=lfs merge=lfs -text
47
+ *.ogg filter=lfs diff=lfs merge=lfs -text
48
+ *.wav filter=lfs diff=lfs merge=lfs -text
49
+ # Image files - uncompressed
50
+ *.bmp filter=lfs diff=lfs merge=lfs -text
51
+ *.gif filter=lfs diff=lfs merge=lfs -text
52
+ *.png filter=lfs diff=lfs merge=lfs -text
53
+ *.tiff filter=lfs diff=lfs merge=lfs -text
54
+ # Image files - compressed
55
+ *.jpg filter=lfs diff=lfs merge=lfs -text
56
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
57
+ *.webp filter=lfs diff=lfs merge=lfs -text
58
+ # Video files - compressed
59
+ *.mp4 filter=lfs diff=lfs merge=lfs -text
60
+ *.webm filter=lfs diff=lfs merge=lfs -text
experiments/task-transfer/provenance/defend_the_line_200ep/.gitattributes ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.avro filter=lfs diff=lfs merge=lfs -text
4
+ *.bin filter=lfs diff=lfs merge=lfs -text
5
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
6
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
7
+ *.ftz filter=lfs diff=lfs merge=lfs -text
8
+ *.gz filter=lfs diff=lfs merge=lfs -text
9
+ *.h5 filter=lfs diff=lfs merge=lfs -text
10
+ *.joblib filter=lfs diff=lfs merge=lfs -text
11
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
12
+ *.lz4 filter=lfs diff=lfs merge=lfs -text
13
+ *.mds filter=lfs diff=lfs merge=lfs -text
14
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
15
+ *.model filter=lfs diff=lfs merge=lfs -text
16
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
17
+ *.npy filter=lfs diff=lfs merge=lfs -text
18
+ *.npz filter=lfs diff=lfs merge=lfs -text
19
+ *.onnx filter=lfs diff=lfs merge=lfs -text
20
+ *.ot filter=lfs diff=lfs merge=lfs -text
21
+ *.parquet filter=lfs diff=lfs merge=lfs -text
22
+ *.pb filter=lfs diff=lfs merge=lfs -text
23
+ *.pickle filter=lfs diff=lfs merge=lfs -text
24
+ *.pkl filter=lfs diff=lfs merge=lfs -text
25
+ *.pt filter=lfs diff=lfs merge=lfs -text
26
+ *.pth filter=lfs diff=lfs merge=lfs -text
27
+ *.rar filter=lfs diff=lfs merge=lfs -text
28
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
29
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
30
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
31
+ *.tar filter=lfs diff=lfs merge=lfs -text
32
+ *.tflite filter=lfs diff=lfs merge=lfs -text
33
+ *.tgz filter=lfs diff=lfs merge=lfs -text
34
+ *.wasm filter=lfs diff=lfs merge=lfs -text
35
+ *.xz filter=lfs diff=lfs merge=lfs -text
36
+ *.zip filter=lfs diff=lfs merge=lfs -text
37
+ *.zst filter=lfs diff=lfs merge=lfs -text
38
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
39
+ # Audio files - uncompressed
40
+ *.pcm filter=lfs diff=lfs merge=lfs -text
41
+ *.sam filter=lfs diff=lfs merge=lfs -text
42
+ *.raw filter=lfs diff=lfs merge=lfs -text
43
+ # Audio files - compressed
44
+ *.aac filter=lfs diff=lfs merge=lfs -text
45
+ *.flac filter=lfs diff=lfs merge=lfs -text
46
+ *.mp3 filter=lfs diff=lfs merge=lfs -text
47
+ *.ogg filter=lfs diff=lfs merge=lfs -text
48
+ *.wav filter=lfs diff=lfs merge=lfs -text
49
+ # Image files - uncompressed
50
+ *.bmp filter=lfs diff=lfs merge=lfs -text
51
+ *.gif filter=lfs diff=lfs merge=lfs -text
52
+ *.png filter=lfs diff=lfs merge=lfs -text
53
+ *.tiff filter=lfs diff=lfs merge=lfs -text
54
+ # Image files - compressed
55
+ *.jpg filter=lfs diff=lfs merge=lfs -text
56
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
57
+ *.webp filter=lfs diff=lfs merge=lfs -text
58
+ # Video files - compressed
59
+ *.mp4 filter=lfs diff=lfs merge=lfs -text
60
+ *.webm filter=lfs diff=lfs merge=lfs -text
experiments/task-transfer/provenance/defend_the_line_200ep/README.md ADDED
@@ -0,0 +1,127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ dataset_info:
3
+ - config_name: defend_the_line_fixed_latency_0_200ep_7k2steps
4
+ features:
5
+ - name: image
6
+ dtype: image
7
+ - name: env_name
8
+ dtype: string
9
+ - name: episode_idx
10
+ dtype: int64
11
+ - name: decision_step
12
+ dtype: int64
13
+ - name: split
14
+ dtype: string
15
+ - name: seed
16
+ dtype: int64
17
+ - name: prompt
18
+ dtype: string
19
+ - name: latency_raw_frames
20
+ dtype: int64
21
+ - name: latency_ms
22
+ dtype: float64
23
+ - name: raw_reward
24
+ dtype: float64
25
+ - name: action_id
26
+ dtype: int64
27
+ - name: action_text
28
+ dtype: string
29
+ splits:
30
+ - name: train
31
+ num_bytes: 895016703
32
+ num_examples: 55907
33
+ - name: val
34
+ num_bytes: 9959107
35
+ num_examples: 626
36
+ download_size: 875557436
37
+ dataset_size: 904975810
38
+ - config_name: defend_the_line_fixed_latency_2_200ep_7k2steps
39
+ features:
40
+ - name: image
41
+ dtype: image
42
+ - name: env_name
43
+ dtype: string
44
+ - name: episode_idx
45
+ dtype: int64
46
+ - name: decision_step
47
+ dtype: int64
48
+ - name: split
49
+ dtype: string
50
+ - name: seed
51
+ dtype: int64
52
+ - name: prompt
53
+ dtype: string
54
+ - name: latency_raw_frames
55
+ dtype: int64
56
+ - name: latency_ms
57
+ dtype: float64
58
+ - name: raw_reward
59
+ dtype: float64
60
+ - name: action_id
61
+ dtype: int64
62
+ - name: action_text
63
+ dtype: string
64
+ splits:
65
+ - name: train
66
+ num_bytes: 1024585898
67
+ num_examples: 63623
68
+ - name: val
69
+ num_bytes: 10401289
70
+ num_examples: 644
71
+ download_size: 1002177725
72
+ dataset_size: 1034987187
73
+ - config_name: defend_the_line_fixed_latency_4_200ep_7k2steps
74
+ features:
75
+ - name: image
76
+ dtype: image
77
+ - name: env_name
78
+ dtype: string
79
+ - name: episode_idx
80
+ dtype: int64
81
+ - name: decision_step
82
+ dtype: int64
83
+ - name: split
84
+ dtype: string
85
+ - name: seed
86
+ dtype: int64
87
+ - name: prompt
88
+ dtype: string
89
+ - name: latency_raw_frames
90
+ dtype: int64
91
+ - name: latency_ms
92
+ dtype: float64
93
+ - name: raw_reward
94
+ dtype: float64
95
+ - name: action_id
96
+ dtype: int64
97
+ - name: action_text
98
+ dtype: string
99
+ splits:
100
+ - name: train
101
+ num_bytes: 966780806
102
+ num_examples: 60249
103
+ - name: val
104
+ num_bytes: 10494921
105
+ num_examples: 657
106
+ download_size: 946114447
107
+ dataset_size: 977275727
108
+ configs:
109
+ - config_name: defend_the_line_fixed_latency_0_200ep_7k2steps
110
+ data_files:
111
+ - split: train
112
+ path: defend_the_line_fixed_latency_0_200ep_7k2steps/train-*
113
+ - split: val
114
+ path: defend_the_line_fixed_latency_0_200ep_7k2steps/val-*
115
+ - config_name: defend_the_line_fixed_latency_2_200ep_7k2steps
116
+ data_files:
117
+ - split: train
118
+ path: defend_the_line_fixed_latency_2_200ep_7k2steps/train-*
119
+ - split: val
120
+ path: defend_the_line_fixed_latency_2_200ep_7k2steps/val-*
121
+ - config_name: defend_the_line_fixed_latency_4_200ep_7k2steps
122
+ data_files:
123
+ - split: train
124
+ path: defend_the_line_fixed_latency_4_200ep_7k2steps/train-*
125
+ - split: val
126
+ path: defend_the_line_fixed_latency_4_200ep_7k2steps/val-*
127
+ ---
latency-aware/asterix/asterix_200ep/asterix_fixed_latency_2_200ep_7k2steps/train-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:048589d78216b1da859011943e19be246dd250f0717bde883f4f988358c9de8f
3
+ size 326368562
latency-aware/asterix/asterix_200ep/asterix_fixed_latency_2_200ep_7k2steps/val-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ee6ab4b71064f29db50e01671940c26555f706870be3b242166899c6880f10b
3
+ size 3319339
latency-aware/asterix/asterix_200ep/asterix_fixed_latency_4_200ep_7k2steps/train-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9617ce0ebda127d70d84db391c2c73ebfebd6dac8cc0ef0f14fda6786aaa9d66
3
+ size 280637541
latency-aware/asterix/asterix_200ep/asterix_fixed_latency_4_200ep_7k2steps/val-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fdbb8f742a72aed8c48180bc7846b69e34b319e0300a792b1f605cf858173c1
3
+ size 2778568
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/filter_report.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_episodes": 1000,
3
+ "accepted_train_episodes": 900,
4
+ "accepted_val_episodes": 100,
5
+ "attempted_episodes": 3204,
6
+ "preset": "clean_v1",
7
+ "rejected_episodes": 2204,
8
+ "train_rows_after_row_filter": 32500,
9
+ "train_rows_before_row_filter": 44188,
10
+ "val_rows_after_row_filter": 4000,
11
+ "val_rows_before_row_filter": 4919
12
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/metadata.json ADDED
@@ -0,0 +1,350 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_labels": [
3
+ "NOOP",
4
+ "ATTACK",
5
+ "MOVE_LEFT",
6
+ "MOVE_LEFT + ATTACK",
7
+ "MOVE_RIGHT",
8
+ "MOVE_RIGHT + ATTACK",
9
+ "MOVE_FORWARD",
10
+ "MOVE_FORWARD + ATTACK",
11
+ "MOVE_FORWARD + MOVE_LEFT",
12
+ "MOVE_FORWARD + MOVE_LEFT + ATTACK",
13
+ "MOVE_FORWARD + MOVE_RIGHT",
14
+ "MOVE_FORWARD + MOVE_RIGHT + ATTACK",
15
+ "MOVE_BACKWARD",
16
+ "MOVE_BACKWARD + ATTACK",
17
+ "MOVE_BACKWARD + MOVE_LEFT",
18
+ "MOVE_BACKWARD + MOVE_LEFT + ATTACK",
19
+ "MOVE_BACKWARD + MOVE_RIGHT",
20
+ "MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
21
+ "TURN_LEFT",
22
+ "TURN_LEFT + ATTACK",
23
+ "TURN_LEFT + MOVE_LEFT",
24
+ "TURN_LEFT + MOVE_LEFT + ATTACK",
25
+ "TURN_LEFT + MOVE_RIGHT",
26
+ "TURN_LEFT + MOVE_RIGHT + ATTACK",
27
+ "TURN_LEFT + MOVE_FORWARD",
28
+ "TURN_LEFT + MOVE_FORWARD + ATTACK",
29
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT",
30
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
31
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT",
32
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
33
+ "TURN_LEFT + MOVE_BACKWARD",
34
+ "TURN_LEFT + MOVE_BACKWARD + ATTACK",
35
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT",
36
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
37
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT",
38
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
39
+ "TURN_RIGHT",
40
+ "TURN_RIGHT + ATTACK",
41
+ "TURN_RIGHT + MOVE_LEFT",
42
+ "TURN_RIGHT + MOVE_LEFT + ATTACK",
43
+ "TURN_RIGHT + MOVE_RIGHT",
44
+ "TURN_RIGHT + MOVE_RIGHT + ATTACK",
45
+ "TURN_RIGHT + MOVE_FORWARD",
46
+ "TURN_RIGHT + MOVE_FORWARD + ATTACK",
47
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT",
48
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
49
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT",
50
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
51
+ "TURN_RIGHT + MOVE_BACKWARD",
52
+ "TURN_RIGHT + MOVE_BACKWARD + ATTACK",
53
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT",
54
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
55
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT",
56
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK"
57
+ ],
58
+ "action_layout": "deadly_corridor_joint_54",
59
+ "base_prompt": "You are playing Deadly Corridor in VizDoom. Your goal is to survive the corridor, fight enemies, and reach the green armor vest at the far end. The available actions are: MOVE_FORWARD, MOVE_BACKWARD, MOVE_LEFT, MOVE_RIGHT, TURN_LEFT, TURN_RIGHT, and ATTACK. Choose the best next action from the current visual observation.",
60
+ "checkpoint_config": {
61
+ "actor_critic_share_weights": true,
62
+ "actor_worker_gpus": [],
63
+ "adam_beta1": 0.9,
64
+ "adam_beta2": 0.999,
65
+ "adam_eps": 1e-06,
66
+ "adaptive_stddev": true,
67
+ "add_latency_info": false,
68
+ "algo": "APPO",
69
+ "async_rl": true,
70
+ "batch_size": 1024,
71
+ "batched_sampling": false,
72
+ "benchmark": false,
73
+ "cli_args": {
74
+ "add_latency_info": false,
75
+ "algo": "APPO",
76
+ "async_rl": true,
77
+ "batch_size": 1024,
78
+ "device": "gpu",
79
+ "disable_reward_scaling": true,
80
+ "env": "doom_deadly_corridor",
81
+ "env_frameskip": 4,
82
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l2_seed0/episode_metrics.jsonl",
83
+ "episodes_per_stage": 10,
84
+ "eval_deterministic": true,
85
+ "eval_env_frameskip": 4,
86
+ "eval_episodes": 20,
87
+ "eval_max_steps": 30000,
88
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
89
+ "eval_raw_reward": true,
90
+ "experiment": "deadly_corridor_sf_same_latency_l2_seed0",
91
+ "experiment_summaries_interval": 1,
92
+ "exploration_loss": "symmetric_kl",
93
+ "exploration_loss_coeff": 0.001,
94
+ "frame_stack": 4,
95
+ "gae_lambda": 0.95,
96
+ "gamma": 0.99,
97
+ "keep_checkpoints": 5,
98
+ "latency_mode": "episode_fixed",
99
+ "latency_sampling_mode": "uniform_all",
100
+ "learning_rate": 0.0001,
101
+ "max_episode_steps": 2100,
102
+ "max_grad_norm": 4.0,
103
+ "mode": "train",
104
+ "normalize_input": true,
105
+ "normalize_returns": true,
106
+ "num_batches_per_epoch": 2,
107
+ "num_envs_per_worker": 4,
108
+ "num_epochs": 2,
109
+ "num_policies": 1,
110
+ "num_workers": 16,
111
+ "ppo_clip_ratio": 0.1,
112
+ "ppo_clip_value": 0.2,
113
+ "queue_latency_values": "2",
114
+ "randomize_within_stage": false,
115
+ "recurrence": 128,
116
+ "res_h": 72,
117
+ "res_w": 128,
118
+ "restart_behavior": "overwrite",
119
+ "rnn_size": 512,
120
+ "rnn_type": "gru",
121
+ "rollout": 128,
122
+ "save_every_sec": 600,
123
+ "seed": 0,
124
+ "stats_avg": 100,
125
+ "train_dir": "results/checkpoints",
126
+ "train_for_env_steps": 50000000,
127
+ "use_rnn": true,
128
+ "value_loss_coeff": 0.5,
129
+ "wandb_group": "deadly_corridor",
130
+ "wandb_job_type": "sample_factory",
131
+ "wandb_project": "latency-sensitive-bench",
132
+ "wide_aspect_ratio": false,
133
+ "with_wandb": true,
134
+ "within_episode_change_interval": 1,
135
+ "within_episode_change_prob": 1.0,
136
+ "worker_num_splits": 2
137
+ },
138
+ "command_line": "--mode train --algo APPO --env doom_deadly_corridor --experiment deadly_corridor_sf_same_latency_l2_seed0 --train_dir results/checkpoints --restart_behavior overwrite --device gpu --seed 0 --episode_metrics_path results/checkpoints/deadly_corridor_sf_same_latency_l2_seed0/episode_metrics.jsonl --train_for_env_steps 50000000 --num_workers 16 --num_envs_per_worker 4 --num_policies 1 --batch_size 1024 --rollout 128 --recurrence 128 --num_epochs 2 --num_batches_per_epoch 2 --worker_num_splits 2 --learning_rate 0.0001 --gamma 0.99 --gae_lambda 0.95 --ppo_clip_ratio 0.1 --ppo_clip_value 0.2 --exploration_loss symmetric_kl --exploration_loss_coeff 0.001 --value_loss_coeff 0.5 --max_grad_norm 4.0 --rnn_type gru --rnn_size 512 --save_every_sec 600 --keep_checkpoints 5 --stats_avg 100 --experiment_summaries_interval 1 --async_rl True --use_rnn True --normalize_returns True --normalize_input True --queue-latency-values 2 --latency-mode episode_fixed --latency-sampling-mode uniform_all --episodes-per-stage 10 --randomize-within-stage False --within-episode-change-interval 1 --within-episode-change-prob 1.0 --add-latency-info False --eval-episodes 20 --eval-queue-latency-values 0,1,2,3,4,5,10,15,40 --eval-max-steps 30000 --eval-deterministic True --eval-raw-reward --with_wandb True --wandb_project latency-sensitive-bench --wandb_group deadly_corridor --wandb_job_type sample_factory --env_frameskip 4 --eval_env_frameskip 4 --res_w 128 --res_h 72 --wide_aspect_ratio False --frame_stack 4 --max_episode_steps 2100 --disable-reward-scaling True",
139
+ "continuous_tanh_scale": 0.0,
140
+ "decoder_mlp_layers": [],
141
+ "decorrelate_envs_on_one_worker": true,
142
+ "decorrelate_experience_max_seconds": 0,
143
+ "default_niceness": 0,
144
+ "device": "gpu",
145
+ "disable_reward_scaling": true,
146
+ "encoder_conv_architecture": "convnet_simple",
147
+ "encoder_conv_mlp_layers": [
148
+ 512
149
+ ],
150
+ "encoder_mlp_layers": [
151
+ 512,
152
+ 512
153
+ ],
154
+ "enjoy_script": null,
155
+ "env": "doom_deadly_corridor",
156
+ "env_frameskip": 4,
157
+ "env_framestack": 1,
158
+ "env_gpu_actions": false,
159
+ "env_gpu_observations": true,
160
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l2_seed0/episode_metrics.jsonl",
161
+ "episodes_per_stage": 10,
162
+ "eval_deterministic": true,
163
+ "eval_env_frameskip": 4,
164
+ "eval_episodes": 20,
165
+ "eval_max_steps": 30000,
166
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
167
+ "eval_raw_reward": true,
168
+ "experiment": "deadly_corridor_sf_same_latency_l2_seed0",
169
+ "experiment_summaries_interval": 1,
170
+ "exploration_loss": "symmetric_kl",
171
+ "exploration_loss_coeff": 0.001,
172
+ "flush_summaries_interval": 30,
173
+ "force_envs_single_thread": false,
174
+ "fps": 35,
175
+ "frame_stack": 4,
176
+ "gae_lambda": 0.95,
177
+ "gamma": 0.99,
178
+ "git_hash": "41bc95edbf670a1190d128da22e9cd3777b87989",
179
+ "git_repo_name": "git@github.com:ZihanWang314/latency-sensitive-bench.git",
180
+ "heartbeat_interval": 20,
181
+ "heartbeat_reporting_interval": 600,
182
+ "help": false,
183
+ "hf_repository": null,
184
+ "initial_stddev": 1.0,
185
+ "keep_checkpoints": 5,
186
+ "kl_loss_coeff": 0.0,
187
+ "latency_mode": "episode_fixed",
188
+ "latency_sampling_mode": "uniform_all",
189
+ "learning_rate": 0.0001,
190
+ "load_checkpoint_kind": "latest",
191
+ "log_to_file": true,
192
+ "lr_adaptive_max": 0.01,
193
+ "lr_adaptive_min": 1e-06,
194
+ "lr_schedule": "constant",
195
+ "lr_schedule_kl_threshold": 0.008,
196
+ "max_episode_steps": 2100,
197
+ "max_grad_norm": 4.0,
198
+ "max_num_episodes": 1000000000.0,
199
+ "max_num_frames": 1000000000.0,
200
+ "max_policy_lag": 1000,
201
+ "mode": "train",
202
+ "no_render": false,
203
+ "nonlinearity": "elu",
204
+ "normalize_input": true,
205
+ "normalize_input_keys": null,
206
+ "normalize_returns": true,
207
+ "num_agents": -1,
208
+ "num_batches_per_epoch": 2,
209
+ "num_batches_to_accumulate": 2,
210
+ "num_bots": -1,
211
+ "num_envs_per_worker": 4,
212
+ "num_epochs": 2,
213
+ "num_humans": 0,
214
+ "num_policies": 1,
215
+ "num_workers": 16,
216
+ "obs_scale": 255.0,
217
+ "obs_subtract_mean": 0.0,
218
+ "optimizer": "adam",
219
+ "output_dir": "outputs/train/deadly_corridor/latency_2/seed_0",
220
+ "pbt_mix_policies_in_one_env": true,
221
+ "pbt_mutation_rate": 0.15,
222
+ "pbt_optimize_gamma": false,
223
+ "pbt_period_env_steps": 5000000,
224
+ "pbt_perturb_max": 1.5,
225
+ "pbt_perturb_min": 1.1,
226
+ "pbt_replace_fraction": 0.3,
227
+ "pbt_replace_reward_gap": 0.1,
228
+ "pbt_replace_reward_gap_absolute": 1e-06,
229
+ "pbt_start_mutation": 20000000,
230
+ "pbt_target_objective": "true_objective",
231
+ "pixel_format": "CHW",
232
+ "policy_index": 0,
233
+ "policy_init_gain": 1.0,
234
+ "policy_initialization": "orthogonal",
235
+ "policy_workers_per_policy": 1,
236
+ "ppo_clip_ratio": 0.1,
237
+ "ppo_clip_value": 0.2,
238
+ "push_to_hub": false,
239
+ "queue_latency_values": "2",
240
+ "randomize_within_stage": false,
241
+ "record_to": null,
242
+ "recurrence": 128,
243
+ "res_h": 72,
244
+ "res_w": 128,
245
+ "restart_behavior": "overwrite",
246
+ "reward_clip": 1000.0,
247
+ "reward_scale": 1.0,
248
+ "rnn_num_layers": 1,
249
+ "rnn_size": 512,
250
+ "rnn_type": "gru",
251
+ "rollout": 128,
252
+ "save_best_after": 100000,
253
+ "save_best_every_sec": 5,
254
+ "save_best_metric": "reward",
255
+ "save_every_sec": 600,
256
+ "save_milestones_sec": -1,
257
+ "save_video": false,
258
+ "seed": 0,
259
+ "serial_mode": false,
260
+ "set_workers_cpu_affinity": true,
261
+ "shuffle_minibatches": false,
262
+ "start_bot_difficulty": null,
263
+ "stats_avg": 100,
264
+ "summaries_use_frameskip": true,
265
+ "timelimit": null,
266
+ "train_dir": "results/checkpoints",
267
+ "train_for_env_steps": 50000000,
268
+ "train_for_seconds": 10000000000,
269
+ "train_script": null,
270
+ "use_env_info_cache": false,
271
+ "use_record_episode_statistics": false,
272
+ "use_rnn": true,
273
+ "value_bootstrap": false,
274
+ "value_loss_coeff": 0.5,
275
+ "video_frames": 1000000000.0,
276
+ "video_name": null,
277
+ "vizdoom_init_lock_max_wait_sec": 60.0,
278
+ "vizdoom_init_lock_timeout_sec": 5.0,
279
+ "vizdoom_init_max_parallel": 10,
280
+ "vizdoom_init_retry_jitter_sec": 0.25,
281
+ "vtrace_c": 1.0,
282
+ "vtrace_rho": 1.0,
283
+ "wandb_group": "deadly_corridor",
284
+ "wandb_job_type": "sample_factory",
285
+ "wandb_project": "latency-sensitive-bench",
286
+ "wandb_tags": [],
287
+ "wandb_unique_id": "deadly-corridor-l2-s0",
288
+ "wandb_user": null,
289
+ "wide_aspect_ratio": false,
290
+ "with_pbt": false,
291
+ "with_vtrace": false,
292
+ "with_wandb": true,
293
+ "within_episode_change_interval": 1,
294
+ "within_episode_change_prob": 1.0,
295
+ "worker_num_splits": 2
296
+ },
297
+ "checkpoint_experiment_dir": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l2_seed0",
298
+ "checkpoint_path": null,
299
+ "checkpoint_train_step": 24420,
300
+ "config_source": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l2_seed0/config.json",
301
+ "env_fps": 35.0,
302
+ "env_name": "deadly_corridor",
303
+ "episode_return_field": "episode_raw_return",
304
+ "filter_preset": "clean_v1",
305
+ "filter_report": {
306
+ "accepted_episodes": 1000,
307
+ "accepted_train_episodes": 900,
308
+ "accepted_val_episodes": 100,
309
+ "attempted_episodes": 3204,
310
+ "preset": "clean_v1",
311
+ "rejected_episodes": 2204,
312
+ "train_rows_after_row_filter": 32500,
313
+ "train_rows_before_row_filter": 44188,
314
+ "val_rows_after_row_filter": 4000,
315
+ "val_rows_before_row_filter": 4919
316
+ },
317
+ "frame_stack": 4,
318
+ "image_resolution": [
319
+ 120,
320
+ 160,
321
+ 3
322
+ ],
323
+ "image_source": "env_raw_rgb",
324
+ "latency_override": {
325
+ "add_latency_info": false,
326
+ "fixed_latency_ms": 228.57142857142858,
327
+ "type": "fixed"
328
+ },
329
+ "latency_source": "legacy_checkpoint_name",
330
+ "max_episodes": 1000,
331
+ "max_steps_per_episode": 3600,
332
+ "num_envs": 2048,
333
+ "obs_fps": 8.75,
334
+ "obs_stride_raw_frames": 4,
335
+ "prompt_source": "latency_bench.data.prompts",
336
+ "prompt_timing_suffix": " Current action latency is {latency_raw_frames} raw frames ({latency_ms:.2f} ms). The environment runs at {env_fps:g} FPS and observations are emitted at {obs_fps:g} FPS. Choose the best next action.",
337
+ "replay_num_envs": 16,
338
+ "reward_field": "raw_reward",
339
+ "reward_semantics": "raw_reward",
340
+ "rows_unit": "decision_step",
341
+ "schema_version": 2,
342
+ "split_seed": 0,
343
+ "total_episodes": 1000,
344
+ "total_rows": 36500,
345
+ "train_episodes": 900,
346
+ "train_rows": 32500,
347
+ "val_episodes": 100,
348
+ "val_fraction": 0.1,
349
+ "val_rows": 4000
350
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/train.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6627ab830bb16f901841b1f3cfa7c4a6c727e323301a079cf7ffe0f1043089d
3
+ size 497929218
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_2_1000ep/val.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:681b9cd8330aefd7bfe74c9dfac4d6737aca4c48a5c54e4e6864a604ec67a0d4
3
+ size 60912654
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/filter_report.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_episodes": 1000,
3
+ "accepted_train_episodes": 900,
4
+ "accepted_val_episodes": 100,
5
+ "attempted_episodes": 11162,
6
+ "preset": "clean_v1",
7
+ "rejected_episodes": 10162,
8
+ "train_rows_after_row_filter": 28737,
9
+ "train_rows_before_row_filter": 44780,
10
+ "val_rows_after_row_filter": 3178,
11
+ "val_rows_before_row_filter": 4925
12
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/metadata.json ADDED
@@ -0,0 +1,350 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_labels": [
3
+ "NOOP",
4
+ "ATTACK",
5
+ "MOVE_LEFT",
6
+ "MOVE_LEFT + ATTACK",
7
+ "MOVE_RIGHT",
8
+ "MOVE_RIGHT + ATTACK",
9
+ "MOVE_FORWARD",
10
+ "MOVE_FORWARD + ATTACK",
11
+ "MOVE_FORWARD + MOVE_LEFT",
12
+ "MOVE_FORWARD + MOVE_LEFT + ATTACK",
13
+ "MOVE_FORWARD + MOVE_RIGHT",
14
+ "MOVE_FORWARD + MOVE_RIGHT + ATTACK",
15
+ "MOVE_BACKWARD",
16
+ "MOVE_BACKWARD + ATTACK",
17
+ "MOVE_BACKWARD + MOVE_LEFT",
18
+ "MOVE_BACKWARD + MOVE_LEFT + ATTACK",
19
+ "MOVE_BACKWARD + MOVE_RIGHT",
20
+ "MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
21
+ "TURN_LEFT",
22
+ "TURN_LEFT + ATTACK",
23
+ "TURN_LEFT + MOVE_LEFT",
24
+ "TURN_LEFT + MOVE_LEFT + ATTACK",
25
+ "TURN_LEFT + MOVE_RIGHT",
26
+ "TURN_LEFT + MOVE_RIGHT + ATTACK",
27
+ "TURN_LEFT + MOVE_FORWARD",
28
+ "TURN_LEFT + MOVE_FORWARD + ATTACK",
29
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT",
30
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
31
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT",
32
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
33
+ "TURN_LEFT + MOVE_BACKWARD",
34
+ "TURN_LEFT + MOVE_BACKWARD + ATTACK",
35
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT",
36
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
37
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT",
38
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
39
+ "TURN_RIGHT",
40
+ "TURN_RIGHT + ATTACK",
41
+ "TURN_RIGHT + MOVE_LEFT",
42
+ "TURN_RIGHT + MOVE_LEFT + ATTACK",
43
+ "TURN_RIGHT + MOVE_RIGHT",
44
+ "TURN_RIGHT + MOVE_RIGHT + ATTACK",
45
+ "TURN_RIGHT + MOVE_FORWARD",
46
+ "TURN_RIGHT + MOVE_FORWARD + ATTACK",
47
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT",
48
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
49
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT",
50
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
51
+ "TURN_RIGHT + MOVE_BACKWARD",
52
+ "TURN_RIGHT + MOVE_BACKWARD + ATTACK",
53
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT",
54
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
55
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT",
56
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK"
57
+ ],
58
+ "action_layout": "deadly_corridor_joint_54",
59
+ "base_prompt": "You are playing Deadly Corridor in VizDoom. Your goal is to survive the corridor, fight enemies, and reach the green armor vest at the far end. The available actions are: MOVE_FORWARD, MOVE_BACKWARD, MOVE_LEFT, MOVE_RIGHT, TURN_LEFT, TURN_RIGHT, and ATTACK. Choose the best next action from the current visual observation.",
60
+ "checkpoint_config": {
61
+ "actor_critic_share_weights": true,
62
+ "actor_worker_gpus": [],
63
+ "adam_beta1": 0.9,
64
+ "adam_beta2": 0.999,
65
+ "adam_eps": 1e-06,
66
+ "adaptive_stddev": true,
67
+ "add_latency_info": false,
68
+ "algo": "APPO",
69
+ "async_rl": true,
70
+ "batch_size": 1024,
71
+ "batched_sampling": false,
72
+ "benchmark": false,
73
+ "cli_args": {
74
+ "add_latency_info": false,
75
+ "algo": "APPO",
76
+ "async_rl": true,
77
+ "batch_size": 1024,
78
+ "device": "gpu",
79
+ "disable_reward_scaling": true,
80
+ "env": "doom_deadly_corridor",
81
+ "env_frameskip": 4,
82
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l4_seed0/episode_metrics.jsonl",
83
+ "episodes_per_stage": 10,
84
+ "eval_deterministic": true,
85
+ "eval_env_frameskip": 4,
86
+ "eval_episodes": 20,
87
+ "eval_max_steps": 30000,
88
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
89
+ "eval_raw_reward": true,
90
+ "experiment": "deadly_corridor_sf_same_latency_l4_seed0",
91
+ "experiment_summaries_interval": 1,
92
+ "exploration_loss": "symmetric_kl",
93
+ "exploration_loss_coeff": 0.001,
94
+ "frame_stack": 4,
95
+ "gae_lambda": 0.95,
96
+ "gamma": 0.99,
97
+ "keep_checkpoints": 5,
98
+ "latency_mode": "episode_fixed",
99
+ "latency_sampling_mode": "uniform_all",
100
+ "learning_rate": 0.0001,
101
+ "max_episode_steps": 2100,
102
+ "max_grad_norm": 4.0,
103
+ "mode": "train",
104
+ "normalize_input": true,
105
+ "normalize_returns": true,
106
+ "num_batches_per_epoch": 2,
107
+ "num_envs_per_worker": 4,
108
+ "num_epochs": 2,
109
+ "num_policies": 1,
110
+ "num_workers": 16,
111
+ "ppo_clip_ratio": 0.1,
112
+ "ppo_clip_value": 0.2,
113
+ "queue_latency_values": "4",
114
+ "randomize_within_stage": false,
115
+ "recurrence": 128,
116
+ "res_h": 72,
117
+ "res_w": 128,
118
+ "restart_behavior": "overwrite",
119
+ "rnn_size": 512,
120
+ "rnn_type": "gru",
121
+ "rollout": 128,
122
+ "save_every_sec": 600,
123
+ "seed": 0,
124
+ "stats_avg": 100,
125
+ "train_dir": "results/checkpoints",
126
+ "train_for_env_steps": 50000000,
127
+ "use_rnn": true,
128
+ "value_loss_coeff": 0.5,
129
+ "wandb_group": "deadly_corridor",
130
+ "wandb_job_type": "sample_factory",
131
+ "wandb_project": "latency-sensitive-bench",
132
+ "wide_aspect_ratio": false,
133
+ "with_wandb": true,
134
+ "within_episode_change_interval": 1,
135
+ "within_episode_change_prob": 1.0,
136
+ "worker_num_splits": 2
137
+ },
138
+ "command_line": "--mode train --algo APPO --env doom_deadly_corridor --experiment deadly_corridor_sf_same_latency_l4_seed0 --train_dir results/checkpoints --restart_behavior overwrite --device gpu --seed 0 --episode_metrics_path results/checkpoints/deadly_corridor_sf_same_latency_l4_seed0/episode_metrics.jsonl --train_for_env_steps 50000000 --num_workers 16 --num_envs_per_worker 4 --num_policies 1 --batch_size 1024 --rollout 128 --recurrence 128 --num_epochs 2 --num_batches_per_epoch 2 --worker_num_splits 2 --learning_rate 0.0001 --gamma 0.99 --gae_lambda 0.95 --ppo_clip_ratio 0.1 --ppo_clip_value 0.2 --exploration_loss symmetric_kl --exploration_loss_coeff 0.001 --value_loss_coeff 0.5 --max_grad_norm 4.0 --rnn_type gru --rnn_size 512 --save_every_sec 600 --keep_checkpoints 5 --stats_avg 100 --experiment_summaries_interval 1 --async_rl True --use_rnn True --normalize_returns True --normalize_input True --queue-latency-values 4 --latency-mode episode_fixed --latency-sampling-mode uniform_all --episodes-per-stage 10 --randomize-within-stage False --within-episode-change-interval 1 --within-episode-change-prob 1.0 --add-latency-info False --eval-episodes 20 --eval-queue-latency-values 0,1,2,3,4,5,10,15,40 --eval-max-steps 30000 --eval-deterministic True --eval-raw-reward --with_wandb True --wandb_project latency-sensitive-bench --wandb_group deadly_corridor --wandb_job_type sample_factory --env_frameskip 4 --eval_env_frameskip 4 --res_w 128 --res_h 72 --wide_aspect_ratio False --frame_stack 4 --max_episode_steps 2100 --disable-reward-scaling True",
139
+ "continuous_tanh_scale": 0.0,
140
+ "decoder_mlp_layers": [],
141
+ "decorrelate_envs_on_one_worker": true,
142
+ "decorrelate_experience_max_seconds": 0,
143
+ "default_niceness": 0,
144
+ "device": "gpu",
145
+ "disable_reward_scaling": true,
146
+ "encoder_conv_architecture": "convnet_simple",
147
+ "encoder_conv_mlp_layers": [
148
+ 512
149
+ ],
150
+ "encoder_mlp_layers": [
151
+ 512,
152
+ 512
153
+ ],
154
+ "enjoy_script": null,
155
+ "env": "doom_deadly_corridor",
156
+ "env_frameskip": 4,
157
+ "env_framestack": 1,
158
+ "env_gpu_actions": false,
159
+ "env_gpu_observations": true,
160
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l4_seed0/episode_metrics.jsonl",
161
+ "episodes_per_stage": 10,
162
+ "eval_deterministic": true,
163
+ "eval_env_frameskip": 4,
164
+ "eval_episodes": 20,
165
+ "eval_max_steps": 30000,
166
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
167
+ "eval_raw_reward": true,
168
+ "experiment": "deadly_corridor_sf_same_latency_l4_seed0",
169
+ "experiment_summaries_interval": 1,
170
+ "exploration_loss": "symmetric_kl",
171
+ "exploration_loss_coeff": 0.001,
172
+ "flush_summaries_interval": 30,
173
+ "force_envs_single_thread": false,
174
+ "fps": 35,
175
+ "frame_stack": 4,
176
+ "gae_lambda": 0.95,
177
+ "gamma": 0.99,
178
+ "git_hash": "41bc95edbf670a1190d128da22e9cd3777b87989",
179
+ "git_repo_name": "git@github.com:ZihanWang314/latency-sensitive-bench.git",
180
+ "heartbeat_interval": 20,
181
+ "heartbeat_reporting_interval": 600,
182
+ "help": false,
183
+ "hf_repository": null,
184
+ "initial_stddev": 1.0,
185
+ "keep_checkpoints": 5,
186
+ "kl_loss_coeff": 0.0,
187
+ "latency_mode": "episode_fixed",
188
+ "latency_sampling_mode": "uniform_all",
189
+ "learning_rate": 0.0001,
190
+ "load_checkpoint_kind": "latest",
191
+ "log_to_file": true,
192
+ "lr_adaptive_max": 0.01,
193
+ "lr_adaptive_min": 1e-06,
194
+ "lr_schedule": "constant",
195
+ "lr_schedule_kl_threshold": 0.008,
196
+ "max_episode_steps": 2100,
197
+ "max_grad_norm": 4.0,
198
+ "max_num_episodes": 1000000000.0,
199
+ "max_num_frames": 1000000000.0,
200
+ "max_policy_lag": 1000,
201
+ "mode": "train",
202
+ "no_render": false,
203
+ "nonlinearity": "elu",
204
+ "normalize_input": true,
205
+ "normalize_input_keys": null,
206
+ "normalize_returns": true,
207
+ "num_agents": -1,
208
+ "num_batches_per_epoch": 2,
209
+ "num_batches_to_accumulate": 2,
210
+ "num_bots": -1,
211
+ "num_envs_per_worker": 4,
212
+ "num_epochs": 2,
213
+ "num_humans": 0,
214
+ "num_policies": 1,
215
+ "num_workers": 16,
216
+ "obs_scale": 255.0,
217
+ "obs_subtract_mean": 0.0,
218
+ "optimizer": "adam",
219
+ "output_dir": "outputs/train/deadly_corridor/latency_4/seed_0",
220
+ "pbt_mix_policies_in_one_env": true,
221
+ "pbt_mutation_rate": 0.15,
222
+ "pbt_optimize_gamma": false,
223
+ "pbt_period_env_steps": 5000000,
224
+ "pbt_perturb_max": 1.5,
225
+ "pbt_perturb_min": 1.1,
226
+ "pbt_replace_fraction": 0.3,
227
+ "pbt_replace_reward_gap": 0.1,
228
+ "pbt_replace_reward_gap_absolute": 1e-06,
229
+ "pbt_start_mutation": 20000000,
230
+ "pbt_target_objective": "true_objective",
231
+ "pixel_format": "CHW",
232
+ "policy_index": 0,
233
+ "policy_init_gain": 1.0,
234
+ "policy_initialization": "orthogonal",
235
+ "policy_workers_per_policy": 1,
236
+ "ppo_clip_ratio": 0.1,
237
+ "ppo_clip_value": 0.2,
238
+ "push_to_hub": false,
239
+ "queue_latency_values": "4",
240
+ "randomize_within_stage": false,
241
+ "record_to": null,
242
+ "recurrence": 128,
243
+ "res_h": 72,
244
+ "res_w": 128,
245
+ "restart_behavior": "overwrite",
246
+ "reward_clip": 1000.0,
247
+ "reward_scale": 1.0,
248
+ "rnn_num_layers": 1,
249
+ "rnn_size": 512,
250
+ "rnn_type": "gru",
251
+ "rollout": 128,
252
+ "save_best_after": 100000,
253
+ "save_best_every_sec": 5,
254
+ "save_best_metric": "reward",
255
+ "save_every_sec": 600,
256
+ "save_milestones_sec": -1,
257
+ "save_video": false,
258
+ "seed": 0,
259
+ "serial_mode": false,
260
+ "set_workers_cpu_affinity": true,
261
+ "shuffle_minibatches": false,
262
+ "start_bot_difficulty": null,
263
+ "stats_avg": 100,
264
+ "summaries_use_frameskip": true,
265
+ "timelimit": null,
266
+ "train_dir": "results/checkpoints",
267
+ "train_for_env_steps": 50000000,
268
+ "train_for_seconds": 10000000000,
269
+ "train_script": null,
270
+ "use_env_info_cache": false,
271
+ "use_record_episode_statistics": false,
272
+ "use_rnn": true,
273
+ "value_bootstrap": false,
274
+ "value_loss_coeff": 0.5,
275
+ "video_frames": 1000000000.0,
276
+ "video_name": null,
277
+ "vizdoom_init_lock_max_wait_sec": 60.0,
278
+ "vizdoom_init_lock_timeout_sec": 5.0,
279
+ "vizdoom_init_max_parallel": 10,
280
+ "vizdoom_init_retry_jitter_sec": 0.25,
281
+ "vtrace_c": 1.0,
282
+ "vtrace_rho": 1.0,
283
+ "wandb_group": "deadly_corridor",
284
+ "wandb_job_type": "sample_factory",
285
+ "wandb_project": "latency-sensitive-bench",
286
+ "wandb_tags": [],
287
+ "wandb_unique_id": "deadly-corridor-l4-s0",
288
+ "wandb_user": null,
289
+ "wide_aspect_ratio": false,
290
+ "with_pbt": false,
291
+ "with_vtrace": false,
292
+ "with_wandb": true,
293
+ "within_episode_change_interval": 1,
294
+ "within_episode_change_prob": 1.0,
295
+ "worker_num_splits": 2
296
+ },
297
+ "checkpoint_experiment_dir": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l4_seed0",
298
+ "checkpoint_path": null,
299
+ "checkpoint_train_step": 24424,
300
+ "config_source": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l4_seed0/config.json",
301
+ "env_fps": 35.0,
302
+ "env_name": "deadly_corridor",
303
+ "episode_return_field": "episode_raw_return",
304
+ "filter_preset": "clean_v1",
305
+ "filter_report": {
306
+ "accepted_episodes": 1000,
307
+ "accepted_train_episodes": 900,
308
+ "accepted_val_episodes": 100,
309
+ "attempted_episodes": 11162,
310
+ "preset": "clean_v1",
311
+ "rejected_episodes": 10162,
312
+ "train_rows_after_row_filter": 28737,
313
+ "train_rows_before_row_filter": 44780,
314
+ "val_rows_after_row_filter": 3178,
315
+ "val_rows_before_row_filter": 4925
316
+ },
317
+ "frame_stack": 4,
318
+ "image_resolution": [
319
+ 120,
320
+ 160,
321
+ 3
322
+ ],
323
+ "image_source": "env_raw_rgb",
324
+ "latency_override": {
325
+ "add_latency_info": false,
326
+ "fixed_latency_ms": 457.14285714285717,
327
+ "type": "fixed"
328
+ },
329
+ "latency_source": "legacy_checkpoint_name",
330
+ "max_episodes": 1000,
331
+ "max_steps_per_episode": 3600,
332
+ "num_envs": 2048,
333
+ "obs_fps": 8.75,
334
+ "obs_stride_raw_frames": 4,
335
+ "prompt_source": "latency_bench.data.prompts",
336
+ "prompt_timing_suffix": " Current action latency is {latency_raw_frames} raw frames ({latency_ms:.2f} ms). The environment runs at {env_fps:g} FPS and observations are emitted at {obs_fps:g} FPS. Choose the best next action.",
337
+ "replay_num_envs": 16,
338
+ "reward_field": "raw_reward",
339
+ "reward_semantics": "raw_reward",
340
+ "rows_unit": "decision_step",
341
+ "schema_version": 2,
342
+ "split_seed": 0,
343
+ "total_episodes": 1000,
344
+ "total_rows": 31915,
345
+ "train_episodes": 900,
346
+ "train_rows": 28737,
347
+ "val_episodes": 100,
348
+ "val_fraction": 0.1,
349
+ "val_rows": 3178
350
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/probe_state.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_specs": [],
3
+ "attempted_episodes": 0,
4
+ "next_attempt_idx": 0,
5
+ "rejected_episodes": 0,
6
+ "runtime_metadata": {
7
+ "checkpoint_train_step": 24424,
8
+ "env_fps": 35.0,
9
+ "frame_stack": 4,
10
+ "obs_fps": 8.75
11
+ }
12
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/train.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ea5cc6ae47203a19b0a00e1ee4d130b7c8a0ee0c93e15240ba4dd75f0575153
3
+ size 451550808
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_4_1000ep/val.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d1183efc614fb8e5769327a8d010a0a09536b9a1b01fe2b249d9f4eb8c32bc64
3
+ size 49818907
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/filter_report.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_episodes": 1000,
3
+ "accepted_train_episodes": 900,
4
+ "accepted_val_episodes": 100,
5
+ "attempted_episodes": 8715,
6
+ "preset": "clean_v1",
7
+ "rejected_episodes": 7715,
8
+ "train_rows_after_row_filter": 34290,
9
+ "train_rows_before_row_filter": 49828,
10
+ "val_rows_after_row_filter": 3890,
11
+ "val_rows_before_row_filter": 5450
12
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/metadata.json ADDED
@@ -0,0 +1,350 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_labels": [
3
+ "NOOP",
4
+ "ATTACK",
5
+ "MOVE_LEFT",
6
+ "MOVE_LEFT + ATTACK",
7
+ "MOVE_RIGHT",
8
+ "MOVE_RIGHT + ATTACK",
9
+ "MOVE_FORWARD",
10
+ "MOVE_FORWARD + ATTACK",
11
+ "MOVE_FORWARD + MOVE_LEFT",
12
+ "MOVE_FORWARD + MOVE_LEFT + ATTACK",
13
+ "MOVE_FORWARD + MOVE_RIGHT",
14
+ "MOVE_FORWARD + MOVE_RIGHT + ATTACK",
15
+ "MOVE_BACKWARD",
16
+ "MOVE_BACKWARD + ATTACK",
17
+ "MOVE_BACKWARD + MOVE_LEFT",
18
+ "MOVE_BACKWARD + MOVE_LEFT + ATTACK",
19
+ "MOVE_BACKWARD + MOVE_RIGHT",
20
+ "MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
21
+ "TURN_LEFT",
22
+ "TURN_LEFT + ATTACK",
23
+ "TURN_LEFT + MOVE_LEFT",
24
+ "TURN_LEFT + MOVE_LEFT + ATTACK",
25
+ "TURN_LEFT + MOVE_RIGHT",
26
+ "TURN_LEFT + MOVE_RIGHT + ATTACK",
27
+ "TURN_LEFT + MOVE_FORWARD",
28
+ "TURN_LEFT + MOVE_FORWARD + ATTACK",
29
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT",
30
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
31
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT",
32
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
33
+ "TURN_LEFT + MOVE_BACKWARD",
34
+ "TURN_LEFT + MOVE_BACKWARD + ATTACK",
35
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT",
36
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
37
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT",
38
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
39
+ "TURN_RIGHT",
40
+ "TURN_RIGHT + ATTACK",
41
+ "TURN_RIGHT + MOVE_LEFT",
42
+ "TURN_RIGHT + MOVE_LEFT + ATTACK",
43
+ "TURN_RIGHT + MOVE_RIGHT",
44
+ "TURN_RIGHT + MOVE_RIGHT + ATTACK",
45
+ "TURN_RIGHT + MOVE_FORWARD",
46
+ "TURN_RIGHT + MOVE_FORWARD + ATTACK",
47
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT",
48
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
49
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT",
50
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
51
+ "TURN_RIGHT + MOVE_BACKWARD",
52
+ "TURN_RIGHT + MOVE_BACKWARD + ATTACK",
53
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT",
54
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
55
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT",
56
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK"
57
+ ],
58
+ "action_layout": "deadly_corridor_joint_54",
59
+ "base_prompt": "You are playing Deadly Corridor in VizDoom. Your goal is to survive the corridor, fight enemies, and reach the green armor vest at the far end. The available actions are: MOVE_FORWARD, MOVE_BACKWARD, MOVE_LEFT, MOVE_RIGHT, TURN_LEFT, TURN_RIGHT, and ATTACK. Choose the best next action from the current visual observation.",
60
+ "checkpoint_config": {
61
+ "actor_critic_share_weights": true,
62
+ "actor_worker_gpus": [],
63
+ "adam_beta1": 0.9,
64
+ "adam_beta2": 0.999,
65
+ "adam_eps": 1e-06,
66
+ "adaptive_stddev": true,
67
+ "add_latency_info": false,
68
+ "algo": "APPO",
69
+ "async_rl": true,
70
+ "batch_size": 1024,
71
+ "batched_sampling": false,
72
+ "benchmark": false,
73
+ "cli_args": {
74
+ "add_latency_info": false,
75
+ "algo": "APPO",
76
+ "async_rl": true,
77
+ "batch_size": 1024,
78
+ "device": "gpu",
79
+ "disable_reward_scaling": true,
80
+ "env": "doom_deadly_corridor",
81
+ "env_frameskip": 4,
82
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l6_seed0/episode_metrics.jsonl",
83
+ "episodes_per_stage": 10,
84
+ "eval_deterministic": true,
85
+ "eval_env_frameskip": 4,
86
+ "eval_episodes": 20,
87
+ "eval_max_steps": 30000,
88
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
89
+ "eval_raw_reward": true,
90
+ "experiment": "deadly_corridor_sf_same_latency_l6_seed0",
91
+ "experiment_summaries_interval": 1,
92
+ "exploration_loss": "symmetric_kl",
93
+ "exploration_loss_coeff": 0.001,
94
+ "frame_stack": 4,
95
+ "gae_lambda": 0.95,
96
+ "gamma": 0.99,
97
+ "keep_checkpoints": 5,
98
+ "latency_mode": "episode_fixed",
99
+ "latency_sampling_mode": "uniform_all",
100
+ "learning_rate": 0.0001,
101
+ "max_episode_steps": 2100,
102
+ "max_grad_norm": 4.0,
103
+ "mode": "train",
104
+ "normalize_input": true,
105
+ "normalize_returns": true,
106
+ "num_batches_per_epoch": 2,
107
+ "num_envs_per_worker": 4,
108
+ "num_epochs": 2,
109
+ "num_policies": 1,
110
+ "num_workers": 16,
111
+ "ppo_clip_ratio": 0.1,
112
+ "ppo_clip_value": 0.2,
113
+ "queue_latency_values": "6",
114
+ "randomize_within_stage": false,
115
+ "recurrence": 128,
116
+ "res_h": 72,
117
+ "res_w": 128,
118
+ "restart_behavior": "overwrite",
119
+ "rnn_size": 512,
120
+ "rnn_type": "gru",
121
+ "rollout": 128,
122
+ "save_every_sec": 600,
123
+ "seed": 0,
124
+ "stats_avg": 100,
125
+ "train_dir": "results/checkpoints",
126
+ "train_for_env_steps": 50000000,
127
+ "use_rnn": true,
128
+ "value_loss_coeff": 0.5,
129
+ "wandb_group": "deadly_corridor",
130
+ "wandb_job_type": "sample_factory",
131
+ "wandb_project": "latency-sensitive-bench",
132
+ "wide_aspect_ratio": false,
133
+ "with_wandb": true,
134
+ "within_episode_change_interval": 1,
135
+ "within_episode_change_prob": 1.0,
136
+ "worker_num_splits": 2
137
+ },
138
+ "command_line": "--mode train --algo APPO --env doom_deadly_corridor --experiment deadly_corridor_sf_same_latency_l6_seed0 --train_dir results/checkpoints --restart_behavior overwrite --device gpu --seed 0 --episode_metrics_path results/checkpoints/deadly_corridor_sf_same_latency_l6_seed0/episode_metrics.jsonl --train_for_env_steps 50000000 --num_workers 16 --num_envs_per_worker 4 --num_policies 1 --batch_size 1024 --rollout 128 --recurrence 128 --num_epochs 2 --num_batches_per_epoch 2 --worker_num_splits 2 --learning_rate 0.0001 --gamma 0.99 --gae_lambda 0.95 --ppo_clip_ratio 0.1 --ppo_clip_value 0.2 --exploration_loss symmetric_kl --exploration_loss_coeff 0.001 --value_loss_coeff 0.5 --max_grad_norm 4.0 --rnn_type gru --rnn_size 512 --save_every_sec 600 --keep_checkpoints 5 --stats_avg 100 --experiment_summaries_interval 1 --async_rl True --use_rnn True --normalize_returns True --normalize_input True --queue-latency-values 6 --latency-mode episode_fixed --latency-sampling-mode uniform_all --episodes-per-stage 10 --randomize-within-stage False --within-episode-change-interval 1 --within-episode-change-prob 1.0 --add-latency-info False --eval-episodes 20 --eval-queue-latency-values 0,1,2,3,4,5,10,15,40 --eval-max-steps 30000 --eval-deterministic True --eval-raw-reward --with_wandb True --wandb_project latency-sensitive-bench --wandb_group deadly_corridor --wandb_job_type sample_factory --env_frameskip 4 --eval_env_frameskip 4 --res_w 128 --res_h 72 --wide_aspect_ratio False --frame_stack 4 --max_episode_steps 2100 --disable-reward-scaling True",
139
+ "continuous_tanh_scale": 0.0,
140
+ "decoder_mlp_layers": [],
141
+ "decorrelate_envs_on_one_worker": true,
142
+ "decorrelate_experience_max_seconds": 0,
143
+ "default_niceness": 0,
144
+ "device": "gpu",
145
+ "disable_reward_scaling": true,
146
+ "encoder_conv_architecture": "convnet_simple",
147
+ "encoder_conv_mlp_layers": [
148
+ 512
149
+ ],
150
+ "encoder_mlp_layers": [
151
+ 512,
152
+ 512
153
+ ],
154
+ "enjoy_script": null,
155
+ "env": "doom_deadly_corridor",
156
+ "env_frameskip": 4,
157
+ "env_framestack": 1,
158
+ "env_gpu_actions": false,
159
+ "env_gpu_observations": true,
160
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l6_seed0/episode_metrics.jsonl",
161
+ "episodes_per_stage": 10,
162
+ "eval_deterministic": true,
163
+ "eval_env_frameskip": 4,
164
+ "eval_episodes": 20,
165
+ "eval_max_steps": 30000,
166
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
167
+ "eval_raw_reward": true,
168
+ "experiment": "deadly_corridor_sf_same_latency_l6_seed0",
169
+ "experiment_summaries_interval": 1,
170
+ "exploration_loss": "symmetric_kl",
171
+ "exploration_loss_coeff": 0.001,
172
+ "flush_summaries_interval": 30,
173
+ "force_envs_single_thread": false,
174
+ "fps": 35,
175
+ "frame_stack": 4,
176
+ "gae_lambda": 0.95,
177
+ "gamma": 0.99,
178
+ "git_hash": "41bc95edbf670a1190d128da22e9cd3777b87989",
179
+ "git_repo_name": "git@github.com:ZihanWang314/latency-sensitive-bench.git",
180
+ "heartbeat_interval": 20,
181
+ "heartbeat_reporting_interval": 600,
182
+ "help": false,
183
+ "hf_repository": null,
184
+ "initial_stddev": 1.0,
185
+ "keep_checkpoints": 5,
186
+ "kl_loss_coeff": 0.0,
187
+ "latency_mode": "episode_fixed",
188
+ "latency_sampling_mode": "uniform_all",
189
+ "learning_rate": 0.0001,
190
+ "load_checkpoint_kind": "latest",
191
+ "log_to_file": true,
192
+ "lr_adaptive_max": 0.01,
193
+ "lr_adaptive_min": 1e-06,
194
+ "lr_schedule": "constant",
195
+ "lr_schedule_kl_threshold": 0.008,
196
+ "max_episode_steps": 2100,
197
+ "max_grad_norm": 4.0,
198
+ "max_num_episodes": 1000000000.0,
199
+ "max_num_frames": 1000000000.0,
200
+ "max_policy_lag": 1000,
201
+ "mode": "train",
202
+ "no_render": false,
203
+ "nonlinearity": "elu",
204
+ "normalize_input": true,
205
+ "normalize_input_keys": null,
206
+ "normalize_returns": true,
207
+ "num_agents": -1,
208
+ "num_batches_per_epoch": 2,
209
+ "num_batches_to_accumulate": 2,
210
+ "num_bots": -1,
211
+ "num_envs_per_worker": 4,
212
+ "num_epochs": 2,
213
+ "num_humans": 0,
214
+ "num_policies": 1,
215
+ "num_workers": 16,
216
+ "obs_scale": 255.0,
217
+ "obs_subtract_mean": 0.0,
218
+ "optimizer": "adam",
219
+ "output_dir": "outputs/train/deadly_corridor/latency_6/seed_0",
220
+ "pbt_mix_policies_in_one_env": true,
221
+ "pbt_mutation_rate": 0.15,
222
+ "pbt_optimize_gamma": false,
223
+ "pbt_period_env_steps": 5000000,
224
+ "pbt_perturb_max": 1.5,
225
+ "pbt_perturb_min": 1.1,
226
+ "pbt_replace_fraction": 0.3,
227
+ "pbt_replace_reward_gap": 0.1,
228
+ "pbt_replace_reward_gap_absolute": 1e-06,
229
+ "pbt_start_mutation": 20000000,
230
+ "pbt_target_objective": "true_objective",
231
+ "pixel_format": "CHW",
232
+ "policy_index": 0,
233
+ "policy_init_gain": 1.0,
234
+ "policy_initialization": "orthogonal",
235
+ "policy_workers_per_policy": 1,
236
+ "ppo_clip_ratio": 0.1,
237
+ "ppo_clip_value": 0.2,
238
+ "push_to_hub": false,
239
+ "queue_latency_values": "6",
240
+ "randomize_within_stage": false,
241
+ "record_to": null,
242
+ "recurrence": 128,
243
+ "res_h": 72,
244
+ "res_w": 128,
245
+ "restart_behavior": "overwrite",
246
+ "reward_clip": 1000.0,
247
+ "reward_scale": 1.0,
248
+ "rnn_num_layers": 1,
249
+ "rnn_size": 512,
250
+ "rnn_type": "gru",
251
+ "rollout": 128,
252
+ "save_best_after": 100000,
253
+ "save_best_every_sec": 5,
254
+ "save_best_metric": "reward",
255
+ "save_every_sec": 600,
256
+ "save_milestones_sec": -1,
257
+ "save_video": false,
258
+ "seed": 0,
259
+ "serial_mode": false,
260
+ "set_workers_cpu_affinity": true,
261
+ "shuffle_minibatches": false,
262
+ "start_bot_difficulty": null,
263
+ "stats_avg": 100,
264
+ "summaries_use_frameskip": true,
265
+ "timelimit": null,
266
+ "train_dir": "results/checkpoints",
267
+ "train_for_env_steps": 50000000,
268
+ "train_for_seconds": 10000000000,
269
+ "train_script": null,
270
+ "use_env_info_cache": false,
271
+ "use_record_episode_statistics": false,
272
+ "use_rnn": true,
273
+ "value_bootstrap": false,
274
+ "value_loss_coeff": 0.5,
275
+ "video_frames": 1000000000.0,
276
+ "video_name": null,
277
+ "vizdoom_init_lock_max_wait_sec": 60.0,
278
+ "vizdoom_init_lock_timeout_sec": 5.0,
279
+ "vizdoom_init_max_parallel": 10,
280
+ "vizdoom_init_retry_jitter_sec": 0.25,
281
+ "vtrace_c": 1.0,
282
+ "vtrace_rho": 1.0,
283
+ "wandb_group": "deadly_corridor",
284
+ "wandb_job_type": "sample_factory",
285
+ "wandb_project": "latency-sensitive-bench",
286
+ "wandb_tags": [],
287
+ "wandb_unique_id": "deadly-corridor-l6-s0",
288
+ "wandb_user": null,
289
+ "wide_aspect_ratio": false,
290
+ "with_pbt": false,
291
+ "with_vtrace": false,
292
+ "with_wandb": true,
293
+ "within_episode_change_interval": 1,
294
+ "within_episode_change_prob": 1.0,
295
+ "worker_num_splits": 2
296
+ },
297
+ "checkpoint_experiment_dir": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l6_seed0",
298
+ "checkpoint_path": null,
299
+ "checkpoint_train_step": 24420,
300
+ "config_source": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l6_seed0/config.json",
301
+ "env_fps": 35.0,
302
+ "env_name": "deadly_corridor",
303
+ "episode_return_field": "episode_raw_return",
304
+ "filter_preset": "clean_v1",
305
+ "filter_report": {
306
+ "accepted_episodes": 1000,
307
+ "accepted_train_episodes": 900,
308
+ "accepted_val_episodes": 100,
309
+ "attempted_episodes": 8715,
310
+ "preset": "clean_v1",
311
+ "rejected_episodes": 7715,
312
+ "train_rows_after_row_filter": 34290,
313
+ "train_rows_before_row_filter": 49828,
314
+ "val_rows_after_row_filter": 3890,
315
+ "val_rows_before_row_filter": 5450
316
+ },
317
+ "frame_stack": 4,
318
+ "image_resolution": [
319
+ 120,
320
+ 160,
321
+ 3
322
+ ],
323
+ "image_source": "env_raw_rgb",
324
+ "latency_override": {
325
+ "add_latency_info": false,
326
+ "fixed_latency_ms": 685.7142857142857,
327
+ "type": "fixed"
328
+ },
329
+ "latency_source": "legacy_checkpoint_name",
330
+ "max_episodes": 1000,
331
+ "max_steps_per_episode": 3600,
332
+ "num_envs": 2048,
333
+ "obs_fps": 8.75,
334
+ "obs_stride_raw_frames": 4,
335
+ "prompt_source": "latency_bench.data.prompts",
336
+ "prompt_timing_suffix": " Current action latency is {latency_raw_frames} raw frames ({latency_ms:.2f} ms). The environment runs at {env_fps:g} FPS and observations are emitted at {obs_fps:g} FPS. Choose the best next action.",
337
+ "replay_num_envs": 16,
338
+ "reward_field": "raw_reward",
339
+ "reward_semantics": "raw_reward",
340
+ "rows_unit": "decision_step",
341
+ "schema_version": 2,
342
+ "split_seed": 0,
343
+ "total_episodes": 1000,
344
+ "total_rows": 38180,
345
+ "train_episodes": 900,
346
+ "train_rows": 34290,
347
+ "val_episodes": 100,
348
+ "val_fraction": 0.1,
349
+ "val_rows": 3890
350
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/train.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bca0ff3cf7a8866cc57c48dbb69f251b273655b1e615a8b9701e0e69a79dcb7
3
+ size 535881935
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_6_1000ep/val.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d043c98ef817648eef1df10f6ce2e9a6a7262fb50575c5238317b5e0ebb87b0f
3
+ size 61270945
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_8_1000ep/filter_report.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_episodes": 1000,
3
+ "accepted_train_episodes": 900,
4
+ "accepted_val_episodes": 100,
5
+ "attempted_episodes": 14133,
6
+ "preset": "clean_v1",
7
+ "rejected_episodes": 13133,
8
+ "train_rows_after_row_filter": 28895,
9
+ "train_rows_before_row_filter": 52991,
10
+ "val_rows_after_row_filter": 3245,
11
+ "val_rows_before_row_filter": 5879
12
+ }
latency-aware/deadly-corridor/deadly_1000ep/deadly_corridor_fix_latency_8_1000ep/metadata.json ADDED
@@ -0,0 +1,350 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_labels": [
3
+ "NOOP",
4
+ "ATTACK",
5
+ "MOVE_LEFT",
6
+ "MOVE_LEFT + ATTACK",
7
+ "MOVE_RIGHT",
8
+ "MOVE_RIGHT + ATTACK",
9
+ "MOVE_FORWARD",
10
+ "MOVE_FORWARD + ATTACK",
11
+ "MOVE_FORWARD + MOVE_LEFT",
12
+ "MOVE_FORWARD + MOVE_LEFT + ATTACK",
13
+ "MOVE_FORWARD + MOVE_RIGHT",
14
+ "MOVE_FORWARD + MOVE_RIGHT + ATTACK",
15
+ "MOVE_BACKWARD",
16
+ "MOVE_BACKWARD + ATTACK",
17
+ "MOVE_BACKWARD + MOVE_LEFT",
18
+ "MOVE_BACKWARD + MOVE_LEFT + ATTACK",
19
+ "MOVE_BACKWARD + MOVE_RIGHT",
20
+ "MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
21
+ "TURN_LEFT",
22
+ "TURN_LEFT + ATTACK",
23
+ "TURN_LEFT + MOVE_LEFT",
24
+ "TURN_LEFT + MOVE_LEFT + ATTACK",
25
+ "TURN_LEFT + MOVE_RIGHT",
26
+ "TURN_LEFT + MOVE_RIGHT + ATTACK",
27
+ "TURN_LEFT + MOVE_FORWARD",
28
+ "TURN_LEFT + MOVE_FORWARD + ATTACK",
29
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT",
30
+ "TURN_LEFT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
31
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT",
32
+ "TURN_LEFT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
33
+ "TURN_LEFT + MOVE_BACKWARD",
34
+ "TURN_LEFT + MOVE_BACKWARD + ATTACK",
35
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT",
36
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
37
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT",
38
+ "TURN_LEFT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK",
39
+ "TURN_RIGHT",
40
+ "TURN_RIGHT + ATTACK",
41
+ "TURN_RIGHT + MOVE_LEFT",
42
+ "TURN_RIGHT + MOVE_LEFT + ATTACK",
43
+ "TURN_RIGHT + MOVE_RIGHT",
44
+ "TURN_RIGHT + MOVE_RIGHT + ATTACK",
45
+ "TURN_RIGHT + MOVE_FORWARD",
46
+ "TURN_RIGHT + MOVE_FORWARD + ATTACK",
47
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT",
48
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_LEFT + ATTACK",
49
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT",
50
+ "TURN_RIGHT + MOVE_FORWARD + MOVE_RIGHT + ATTACK",
51
+ "TURN_RIGHT + MOVE_BACKWARD",
52
+ "TURN_RIGHT + MOVE_BACKWARD + ATTACK",
53
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT",
54
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_LEFT + ATTACK",
55
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT",
56
+ "TURN_RIGHT + MOVE_BACKWARD + MOVE_RIGHT + ATTACK"
57
+ ],
58
+ "action_layout": "deadly_corridor_joint_54",
59
+ "base_prompt": "You are playing Deadly Corridor in VizDoom. Your goal is to survive the corridor, fight enemies, and reach the green armor vest at the far end. The available actions are: MOVE_FORWARD, MOVE_BACKWARD, MOVE_LEFT, MOVE_RIGHT, TURN_LEFT, TURN_RIGHT, and ATTACK. Choose the best next action from the current visual observation.",
60
+ "checkpoint_config": {
61
+ "actor_critic_share_weights": true,
62
+ "actor_worker_gpus": [],
63
+ "adam_beta1": 0.9,
64
+ "adam_beta2": 0.999,
65
+ "adam_eps": 1e-06,
66
+ "adaptive_stddev": true,
67
+ "add_latency_info": false,
68
+ "algo": "APPO",
69
+ "async_rl": true,
70
+ "batch_size": 1024,
71
+ "batched_sampling": false,
72
+ "benchmark": false,
73
+ "cli_args": {
74
+ "add_latency_info": false,
75
+ "algo": "APPO",
76
+ "async_rl": true,
77
+ "batch_size": 1024,
78
+ "device": "gpu",
79
+ "disable_reward_scaling": true,
80
+ "env": "doom_deadly_corridor",
81
+ "env_frameskip": 4,
82
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l8_seed0/episode_metrics.jsonl",
83
+ "episodes_per_stage": 10,
84
+ "eval_deterministic": true,
85
+ "eval_env_frameskip": 4,
86
+ "eval_episodes": 20,
87
+ "eval_max_steps": 30000,
88
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
89
+ "eval_raw_reward": true,
90
+ "experiment": "deadly_corridor_sf_same_latency_l8_seed0",
91
+ "experiment_summaries_interval": 1,
92
+ "exploration_loss": "symmetric_kl",
93
+ "exploration_loss_coeff": 0.001,
94
+ "frame_stack": 4,
95
+ "gae_lambda": 0.95,
96
+ "gamma": 0.99,
97
+ "keep_checkpoints": 5,
98
+ "latency_mode": "episode_fixed",
99
+ "latency_sampling_mode": "uniform_all",
100
+ "learning_rate": 0.0001,
101
+ "max_episode_steps": 2100,
102
+ "max_grad_norm": 4.0,
103
+ "mode": "train",
104
+ "normalize_input": true,
105
+ "normalize_returns": true,
106
+ "num_batches_per_epoch": 2,
107
+ "num_envs_per_worker": 4,
108
+ "num_epochs": 2,
109
+ "num_policies": 1,
110
+ "num_workers": 16,
111
+ "ppo_clip_ratio": 0.1,
112
+ "ppo_clip_value": 0.2,
113
+ "queue_latency_values": "8",
114
+ "randomize_within_stage": false,
115
+ "recurrence": 128,
116
+ "res_h": 72,
117
+ "res_w": 128,
118
+ "restart_behavior": "overwrite",
119
+ "rnn_size": 512,
120
+ "rnn_type": "gru",
121
+ "rollout": 128,
122
+ "save_every_sec": 600,
123
+ "seed": 0,
124
+ "stats_avg": 100,
125
+ "train_dir": "results/checkpoints",
126
+ "train_for_env_steps": 50000000,
127
+ "use_rnn": true,
128
+ "value_loss_coeff": 0.5,
129
+ "wandb_group": "deadly_corridor",
130
+ "wandb_job_type": "sample_factory",
131
+ "wandb_project": "latency-sensitive-bench",
132
+ "wide_aspect_ratio": false,
133
+ "with_wandb": true,
134
+ "within_episode_change_interval": 1,
135
+ "within_episode_change_prob": 1.0,
136
+ "worker_num_splits": 2
137
+ },
138
+ "command_line": "--mode train --algo APPO --env doom_deadly_corridor --experiment deadly_corridor_sf_same_latency_l8_seed0 --train_dir results/checkpoints --restart_behavior overwrite --device gpu --seed 0 --episode_metrics_path results/checkpoints/deadly_corridor_sf_same_latency_l8_seed0/episode_metrics.jsonl --train_for_env_steps 50000000 --num_workers 16 --num_envs_per_worker 4 --num_policies 1 --batch_size 1024 --rollout 128 --recurrence 128 --num_epochs 2 --num_batches_per_epoch 2 --worker_num_splits 2 --learning_rate 0.0001 --gamma 0.99 --gae_lambda 0.95 --ppo_clip_ratio 0.1 --ppo_clip_value 0.2 --exploration_loss symmetric_kl --exploration_loss_coeff 0.001 --value_loss_coeff 0.5 --max_grad_norm 4.0 --rnn_type gru --rnn_size 512 --save_every_sec 600 --keep_checkpoints 5 --stats_avg 100 --experiment_summaries_interval 1 --async_rl True --use_rnn True --normalize_returns True --normalize_input True --queue-latency-values 8 --latency-mode episode_fixed --latency-sampling-mode uniform_all --episodes-per-stage 10 --randomize-within-stage False --within-episode-change-interval 1 --within-episode-change-prob 1.0 --add-latency-info False --eval-episodes 20 --eval-queue-latency-values 0,1,2,3,4,5,10,15,40 --eval-max-steps 30000 --eval-deterministic True --eval-raw-reward --with_wandb True --wandb_project latency-sensitive-bench --wandb_group deadly_corridor --wandb_job_type sample_factory --env_frameskip 4 --eval_env_frameskip 4 --res_w 128 --res_h 72 --wide_aspect_ratio False --frame_stack 4 --max_episode_steps 2100 --disable-reward-scaling True",
139
+ "continuous_tanh_scale": 0.0,
140
+ "decoder_mlp_layers": [],
141
+ "decorrelate_envs_on_one_worker": true,
142
+ "decorrelate_experience_max_seconds": 0,
143
+ "default_niceness": 0,
144
+ "device": "gpu",
145
+ "disable_reward_scaling": true,
146
+ "encoder_conv_architecture": "convnet_simple",
147
+ "encoder_conv_mlp_layers": [
148
+ 512
149
+ ],
150
+ "encoder_mlp_layers": [
151
+ 512,
152
+ 512
153
+ ],
154
+ "enjoy_script": null,
155
+ "env": "doom_deadly_corridor",
156
+ "env_frameskip": 4,
157
+ "env_framestack": 1,
158
+ "env_gpu_actions": false,
159
+ "env_gpu_observations": true,
160
+ "episode_metrics_path": "results/checkpoints/deadly_corridor_sf_same_latency_l8_seed0/episode_metrics.jsonl",
161
+ "episodes_per_stage": 10,
162
+ "eval_deterministic": true,
163
+ "eval_env_frameskip": 4,
164
+ "eval_episodes": 20,
165
+ "eval_max_steps": 30000,
166
+ "eval_queue_latency_values": "0,1,2,3,4,5,10,15,40",
167
+ "eval_raw_reward": true,
168
+ "experiment": "deadly_corridor_sf_same_latency_l8_seed0",
169
+ "experiment_summaries_interval": 1,
170
+ "exploration_loss": "symmetric_kl",
171
+ "exploration_loss_coeff": 0.001,
172
+ "flush_summaries_interval": 30,
173
+ "force_envs_single_thread": false,
174
+ "fps": 35,
175
+ "frame_stack": 4,
176
+ "gae_lambda": 0.95,
177
+ "gamma": 0.99,
178
+ "git_hash": "41bc95edbf670a1190d128da22e9cd3777b87989",
179
+ "git_repo_name": "git@github.com:ZihanWang314/latency-sensitive-bench.git",
180
+ "heartbeat_interval": 20,
181
+ "heartbeat_reporting_interval": 600,
182
+ "help": false,
183
+ "hf_repository": null,
184
+ "initial_stddev": 1.0,
185
+ "keep_checkpoints": 5,
186
+ "kl_loss_coeff": 0.0,
187
+ "latency_mode": "episode_fixed",
188
+ "latency_sampling_mode": "uniform_all",
189
+ "learning_rate": 0.0001,
190
+ "load_checkpoint_kind": "latest",
191
+ "log_to_file": true,
192
+ "lr_adaptive_max": 0.01,
193
+ "lr_adaptive_min": 1e-06,
194
+ "lr_schedule": "constant",
195
+ "lr_schedule_kl_threshold": 0.008,
196
+ "max_episode_steps": 2100,
197
+ "max_grad_norm": 4.0,
198
+ "max_num_episodes": 1000000000.0,
199
+ "max_num_frames": 1000000000.0,
200
+ "max_policy_lag": 1000,
201
+ "mode": "train",
202
+ "no_render": false,
203
+ "nonlinearity": "elu",
204
+ "normalize_input": true,
205
+ "normalize_input_keys": null,
206
+ "normalize_returns": true,
207
+ "num_agents": -1,
208
+ "num_batches_per_epoch": 2,
209
+ "num_batches_to_accumulate": 2,
210
+ "num_bots": -1,
211
+ "num_envs_per_worker": 4,
212
+ "num_epochs": 2,
213
+ "num_humans": 0,
214
+ "num_policies": 1,
215
+ "num_workers": 16,
216
+ "obs_scale": 255.0,
217
+ "obs_subtract_mean": 0.0,
218
+ "optimizer": "adam",
219
+ "output_dir": "outputs/train/deadly_corridor/latency_8/seed_0",
220
+ "pbt_mix_policies_in_one_env": true,
221
+ "pbt_mutation_rate": 0.15,
222
+ "pbt_optimize_gamma": false,
223
+ "pbt_period_env_steps": 5000000,
224
+ "pbt_perturb_max": 1.5,
225
+ "pbt_perturb_min": 1.1,
226
+ "pbt_replace_fraction": 0.3,
227
+ "pbt_replace_reward_gap": 0.1,
228
+ "pbt_replace_reward_gap_absolute": 1e-06,
229
+ "pbt_start_mutation": 20000000,
230
+ "pbt_target_objective": "true_objective",
231
+ "pixel_format": "CHW",
232
+ "policy_index": 0,
233
+ "policy_init_gain": 1.0,
234
+ "policy_initialization": "orthogonal",
235
+ "policy_workers_per_policy": 1,
236
+ "ppo_clip_ratio": 0.1,
237
+ "ppo_clip_value": 0.2,
238
+ "push_to_hub": false,
239
+ "queue_latency_values": "8",
240
+ "randomize_within_stage": false,
241
+ "record_to": null,
242
+ "recurrence": 128,
243
+ "res_h": 72,
244
+ "res_w": 128,
245
+ "restart_behavior": "overwrite",
246
+ "reward_clip": 1000.0,
247
+ "reward_scale": 1.0,
248
+ "rnn_num_layers": 1,
249
+ "rnn_size": 512,
250
+ "rnn_type": "gru",
251
+ "rollout": 128,
252
+ "save_best_after": 100000,
253
+ "save_best_every_sec": 5,
254
+ "save_best_metric": "reward",
255
+ "save_every_sec": 600,
256
+ "save_milestones_sec": -1,
257
+ "save_video": false,
258
+ "seed": 0,
259
+ "serial_mode": false,
260
+ "set_workers_cpu_affinity": true,
261
+ "shuffle_minibatches": false,
262
+ "start_bot_difficulty": null,
263
+ "stats_avg": 100,
264
+ "summaries_use_frameskip": true,
265
+ "timelimit": null,
266
+ "train_dir": "results/checkpoints",
267
+ "train_for_env_steps": 50000000,
268
+ "train_for_seconds": 10000000000,
269
+ "train_script": null,
270
+ "use_env_info_cache": false,
271
+ "use_record_episode_statistics": false,
272
+ "use_rnn": true,
273
+ "value_bootstrap": false,
274
+ "value_loss_coeff": 0.5,
275
+ "video_frames": 1000000000.0,
276
+ "video_name": null,
277
+ "vizdoom_init_lock_max_wait_sec": 60.0,
278
+ "vizdoom_init_lock_timeout_sec": 5.0,
279
+ "vizdoom_init_max_parallel": 10,
280
+ "vizdoom_init_retry_jitter_sec": 0.25,
281
+ "vtrace_c": 1.0,
282
+ "vtrace_rho": 1.0,
283
+ "wandb_group": "deadly_corridor",
284
+ "wandb_job_type": "sample_factory",
285
+ "wandb_project": "latency-sensitive-bench",
286
+ "wandb_tags": [],
287
+ "wandb_unique_id": "deadly-corridor-l8-s0",
288
+ "wandb_user": null,
289
+ "wide_aspect_ratio": false,
290
+ "with_pbt": false,
291
+ "with_vtrace": false,
292
+ "with_wandb": true,
293
+ "within_episode_change_interval": 1,
294
+ "within_episode_change_prob": 1.0,
295
+ "worker_num_splits": 2
296
+ },
297
+ "checkpoint_experiment_dir": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l8_seed0",
298
+ "checkpoint_path": null,
299
+ "checkpoint_train_step": 24420,
300
+ "config_source": "/work/hdd/bfea/tchafekar/small_model_checkpoints/checkpoints_old/deadly_corridor_sf_same_latency_l8_seed0/config.json",
301
+ "env_fps": 35.0,
302
+ "env_name": "deadly_corridor",
303
+ "episode_return_field": "episode_raw_return",
304
+ "filter_preset": "clean_v1",
305
+ "filter_report": {
306
+ "accepted_episodes": 1000,
307
+ "accepted_train_episodes": 900,
308
+ "accepted_val_episodes": 100,
309
+ "attempted_episodes": 14133,
310
+ "preset": "clean_v1",
311
+ "rejected_episodes": 13133,
312
+ "train_rows_after_row_filter": 28895,
313
+ "train_rows_before_row_filter": 52991,
314
+ "val_rows_after_row_filter": 3245,
315
+ "val_rows_before_row_filter": 5879
316
+ },
317
+ "frame_stack": 4,
318
+ "image_resolution": [
319
+ 120,
320
+ 160,
321
+ 3
322
+ ],
323
+ "image_source": "env_raw_rgb",
324
+ "latency_override": {
325
+ "add_latency_info": false,
326
+ "fixed_latency_ms": 914.2857142857143,
327
+ "type": "fixed"
328
+ },
329
+ "latency_source": "legacy_checkpoint_name",
330
+ "max_episodes": 1000,
331
+ "max_steps_per_episode": 3600,
332
+ "num_envs": 2048,
333
+ "obs_fps": 8.75,
334
+ "obs_stride_raw_frames": 4,
335
+ "prompt_source": "latency_bench.data.prompts",
336
+ "prompt_timing_suffix": " Current action latency is {latency_raw_frames} raw frames ({latency_ms:.2f} ms). The environment runs at {env_fps:g} FPS and observations are emitted at {obs_fps:g} FPS. Choose the best next action.",
337
+ "replay_num_envs": 16,
338
+ "reward_field": "raw_reward",
339
+ "reward_semantics": "raw_reward",
340
+ "rows_unit": "decision_step",
341
+ "schema_version": 2,
342
+ "split_seed": 0,
343
+ "total_episodes": 1000,
344
+ "total_rows": 32140,
345
+ "train_episodes": 900,
346
+ "train_rows": 28895,
347
+ "val_episodes": 100,
348
+ "val_fraction": 0.1,
349
+ "val_rows": 3245
350
+ }