| N ←number of rollouts Nopt ← number of optimization steps |
| K←length of rollout |
| M ← number of initial steps for initializing observation normalization t=0 |
| Sample state So ~ po(so) for m = 1 to M do |
| sample at ~ Uniform(at) |
| sample St+1 ~ p(St+1lst, at) |
| Update observation normalization parameters using St+1 |
| t+=1 |
| end for |
| fori=1 to N do |
| for j = 1 to K do |
|
| sample at ~ π(at|St) |
| sample St+1,et ~p(St+1,et|St, at) |
| calculate intrinsic reward it = |lf(St+1) - f(St+1)ll² |
| add St, St+1,at, et,it to optimization batch Bi |
| Update running estimate of reward standard deviation using it |
| t+=1 |
| end for |
| Normalize the intrinsic rewards contained in Bi |
| Calculate returns R1,i and advantages A1,i for intrinsic reward |
| Calculate returns RE,i and advantages AE,i for extrinsic reward |
|
| Calculate combined advantages Ai = A1,i + AE,i |
| Update observation normalization parameters using Bi |
| for j = 1 to Nopt do |
| optimize 0π wrt PPO loss on batch Bi,Ri,Ai using Adam |
|
| optimize 0f wrt distillation loss on Bi using Adam |
|
|
|
| end for |
|
|
|
|
| end for |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
+
+# A.3 PREPROCESSING DETAILS
+
+Table 1 contains details of how we preprocessed the environment for our experiments. We followed the recommendations in Machado et al. (2017) in using sticky actions in order to make the environments non-deterministic so that memorization of action sequences is not possible. In Table 2 we show additional preprocessing details for the policy and value networks. In Table 3 we show additional preprocessing details for the predictor and target networks.
+
+Table 1: Preprocessing details for the environments for all experiments.
+
+| N ←number of rollouts Nopt ← number of optimization steps |
| K←length of rollout |
| M ← number of initial steps for initializing observation normalization t=0 |
| Sample state So ~ po(so) for m = 1 to M do |
| sample at ~ Uniform(at) |
| sample St+1 ~ p(St+1lst, at) |
| Update observation normalization parameters using St+1 |
| t+=1 |
| end for |
| fori=1 to N do |
| for j = 1 to K do |
|
| sample at ~ π(at|St) |
| sample St+1,et ~p(St+1,et|St, at) |
| calculate intrinsic reward it = |lf(St+1) - f(St+1)ll² |
| add St, St+1,at, et,it to optimization batch Bi |
| Update running estimate of reward standard deviation using it |
| t+=1 |
| end for |
| Normalize the intrinsic rewards contained in Bi |
| Calculate returns R1,i and advantages A1,i for intrinsic reward |
| Calculate returns RE,i and advantages AE,i for extrinsic reward |
|
| Calculate combined advantages Ai = A1,i + AE,i |
| Update observation normalization parameters using Bi |
| for j = 1 to Nopt do |
| optimize 0π wrt PPO loss on batch Bi,Ri,Ai using Adam |
|
| optimize 0f wrt distillation loss on Bi using Adam |
|
|
|
| end for |
|
|
|
|
| end for |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
",
+ "bbox": [
+ 179,
+ 117,
+ 818,
+ 559
+ ],
+ "page_idx": 13
+ },
+ {
+ "type": "text",
+ "text": "A.3 PREPROCESSING DETAILS ",
+ "text_level": 1,
+ "bbox": [
+ 176,
+ 595,
+ 397,
+ 609
+ ],
+ "page_idx": 13
+ },
+ {
+ "type": "text",
+ "text": "Table 1 contains details of how we preprocessed the environment for our experiments. We followed the recommendations in Machado et al. (2017) in using sticky actions in order to make the environments non-deterministic so that memorization of action sequences is not possible. In Table 2 we show additional preprocessing details for the policy and value networks. In Table 3 we show additional preprocessing details for the predictor and target networks. ",
+ "bbox": [
+ 173,
+ 626,
+ 825,
+ 696
+ ],
+ "page_idx": 13
+ },
+ {
+ "type": "table",
+ "img_path": "images/06a0a467edba5c6b0b8bfc674d061963b204208f8d6233d3e58968002664507b.jpg",
+ "table_caption": [
+ "Table 1: Preprocessing details for the environments for all experiments. "
+ ],
+ "table_footnote": [],
+ "table_body": "| N ←number of rollouts Nopt ← number of optimization steps |
| K←length of rollout |
| M ← number of initial steps for initializing observation normalization t=0 |
| Sample state So ~ po(so) for m = 1 to M do |
| sample at ~ Uniform(at) |
| sample St+1 ~ p(St+1lst, at) |
| Update observation normalization parameters using St+1 |
| t+=1 |
| end for |
| fori=1 to N do |
| for j = 1 to K do |
|
| sample at ~ π(at|St) |
| sample St+1,et ~p(St+1,et|St, at) |
| calculate intrinsic reward it = |lf(St+1) - f(St+1)ll² |
| add St, St+1,at, et,it to optimization batch Bi |
| Update running estimate of reward standard deviation using it |
| t+=1 |
| end for |
| Normalize the intrinsic rewards contained in Bi |
| Calculate returns R1,i and advantages A1,i for intrinsic reward |
| Calculate returns RE,i and advantages AE,i for extrinsic reward |
|
| Calculate combined advantages Ai = A1,i + AE,i |
| Update observation normalization parameters using Bi |
| for j = 1 to Nopt do |
| optimize 0π wrt PPO loss on batch Bi,Ri,Ai using Adam |
|
| optimize 0f wrt distillation loss on Bi using Adam |
|
|
|
| end for |
|
|
|
|
| end for |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
",
+ "type": "table",
+ "image_path": "591c3503319b8f1bc4963d0f5d81b3980d44fad760389c28a54edeacdcf97fcc.jpg"
+ }
+ ]
+ }
+ ],
+ "index": 2,
+ "virtual_lines": [
+ {
+ "bbox": [
+ 110,
+ 93,
+ 501,
+ 209.66666666666669
+ ],
+ "spans": [],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 110,
+ 209.66666666666669,
+ 501,
+ 326.33333333333337
+ ],
+ "spans": [],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 110,
+ 326.33333333333337,
+ 501,
+ 443.00000000000006
+ ],
+ "spans": [],
+ "index": 3
+ }
+ ]
+ }
+ ],
+ "index": 2
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 108,
+ 472,
+ 243,
+ 483
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 471,
+ 244,
+ 485
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 471,
+ 244,
+ 485
+ ],
+ "score": 1.0,
+ "content": "A.3 PREPROCESSING DETAILS",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 4
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 496,
+ 505,
+ 552
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 496,
+ 505,
+ 509
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 496,
+ 505,
+ 509
+ ],
+ "score": 1.0,
+ "content": "Table 1 contains details of how we preprocessed the environment for our experiments. We followed the",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 105,
+ 507,
+ 505,
+ 520
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 507,
+ 505,
+ 520
+ ],
+ "score": 1.0,
+ "content": "recommendations in Machado et al. (2017) in using sticky actions in order to make the environments",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 106,
+ 518,
+ 504,
+ 530
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 518,
+ 504,
+ 530
+ ],
+ "score": 1.0,
+ "content": "non-deterministic so that memorization of action sequences is not possible. In Table 2 we show",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 106,
+ 529,
+ 505,
+ 541
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 529,
+ 505,
+ 541
+ ],
+ "score": 1.0,
+ "content": "additional preprocessing details for the policy and value networks. In Table 3 we show additional",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 105,
+ 541,
+ 343,
+ 553
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 541,
+ 343,
+ 553
+ ],
+ "score": 1.0,
+ "content": "preprocessing details for the predictor and target networks.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ }
+ ],
+ "index": 7
+ },
+ {
+ "type": "table",
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "blocks": [
+ {
+ "type": "table_body",
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "group_id": 0,
+ "lines": [
+ {
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "score": 0.975,
+ "html": "| N ←number of rollouts Nopt ← number of optimization steps |
| K←length of rollout |
| M ← number of initial steps for initializing observation normalization t=0 |
| Sample state So ~ po(so) for m = 1 to M do |
| sample at ~ Uniform(at) |
| sample St+1 ~ p(St+1lst, at) |
| Update observation normalization parameters using St+1 |
| t+=1 |
| end for |
| fori=1 to N do |
| for j = 1 to K do |
|
| sample at ~ π(at|St) |
| sample St+1,et ~p(St+1,et|St, at) |
| calculate intrinsic reward it = |lf(St+1) - f(St+1)ll² |
| add St, St+1,at, et,it to optimization batch Bi |
| Update running estimate of reward standard deviation using it |
| t+=1 |
| end for |
| Normalize the intrinsic rewards contained in Bi |
| Calculate returns R1,i and advantages A1,i for intrinsic reward |
| Calculate returns RE,i and advantages AE,i for extrinsic reward |
|
| Calculate combined advantages Ai = A1,i + AE,i |
| Update observation normalization parameters using Bi |
| for j = 1 to Nopt do |
| optimize 0π wrt PPO loss on batch Bi,Ri,Ai using Adam |
|
| optimize 0f wrt distillation loss on Bi using Adam |
|
|
|
| end for |
|
|
|
|
| end for |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
",
+ "type": "table",
+ "image_path": "591c3503319b8f1bc4963d0f5d81b3980d44fad760389c28a54edeacdcf97fcc.jpg"
+ }
+ ]
+ }
+ ],
+ "index": 2,
+ "virtual_lines": [
+ {
+ "bbox": [
+ 110,
+ 93,
+ 501,
+ 209.66666666666669
+ ],
+ "spans": [],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 110,
+ 209.66666666666669,
+ 501,
+ 326.33333333333337
+ ],
+ "spans": [],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 110,
+ 326.33333333333337,
+ 501,
+ 443.00000000000006
+ ],
+ "spans": [],
+ "index": 3
+ }
+ ]
+ }
+ ],
+ "index": 2
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 108,
+ 472,
+ 243,
+ 483
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 471,
+ 244,
+ 485
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 471,
+ 244,
+ 485
+ ],
+ "score": 1.0,
+ "content": "A.3 PREPROCESSING DETAILS",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 4
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 496,
+ 505,
+ 552
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 496,
+ 505,
+ 509
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 496,
+ 505,
+ 509
+ ],
+ "score": 1.0,
+ "content": "Table 1 contains details of how we preprocessed the environment for our experiments. We followed the",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 105,
+ 507,
+ 505,
+ 520
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 507,
+ 505,
+ 520
+ ],
+ "score": 1.0,
+ "content": "recommendations in Machado et al. (2017) in using sticky actions in order to make the environments",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 106,
+ 518,
+ 504,
+ 530
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 518,
+ 504,
+ 530
+ ],
+ "score": 1.0,
+ "content": "non-deterministic so that memorization of action sequences is not possible. In Table 2 we show",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 106,
+ 529,
+ 505,
+ 541
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 529,
+ 505,
+ 541
+ ],
+ "score": 1.0,
+ "content": "additional preprocessing details for the policy and value networks. In Table 3 we show additional",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 105,
+ 541,
+ 343,
+ 553
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 541,
+ 343,
+ 553
+ ],
+ "score": 1.0,
+ "content": "preprocessing details for the predictor and target networks.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ }
+ ],
+ "index": 7,
+ "bbox_fs": [
+ 105,
+ 496,
+ 505,
+ 553
+ ]
+ },
+ {
+ "type": "table",
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "blocks": [
+ {
+ "type": "table_body",
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "group_id": 0,
+ "lines": [
+ {
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 223,
+ 569,
+ 388,
+ 687
+ ],
+ "score": 0.975,
+ "html": "| N ←number of rollouts Nopt ← number of optimization steps |
| K←length of rollout |
| M ← number of initial steps for initializing observation normalization t=0 |
| Sample state So ~ po(so) for m = 1 to M do |
| sample at ~ Uniform(at) |
| sample St+1 ~ p(St+1lst, at) |
| Update observation normalization parameters using St+1 |
| t+=1 |
| end for |
| fori=1 to N do |
| for j = 1 to K do |
|
| sample at ~ π(at|St) |
| sample St+1,et ~p(St+1,et|St, at) |
| calculate intrinsic reward it = |lf(St+1) - f(St+1)ll² |
| add St, St+1,at, et,it to optimization batch Bi |
| Update running estimate of reward standard deviation using it |
| t+=1 |
| end for |
| Normalize the intrinsic rewards contained in Bi |
| Calculate returns R1,i and advantages A1,i for intrinsic reward |
| Calculate returns RE,i and advantages AE,i for extrinsic reward |
|
| Calculate combined advantages Ai = A1,i + AE,i |
| Update observation normalization parameters using Bi |
| for j = 1 to Nopt do |
| optimize 0π wrt PPO loss on batch Bi,Ri,Ai using Adam |
|
| optimize 0f wrt distillation loss on Bi using Adam |
|
|
|
| end for |
|
|
|
|
| end for |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
"
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 298.0,
+ 73.0,
+ 858.0,
+ 73.0,
+ 858.0,
+ 108.0,
+ 298.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1311.0,
+ 679.0,
+ 1311.0,
+ 679.0,
+ 1348.0,
+ 295.0,
+ 1348.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 831.0,
+ 2084.0,
+ 869.0,
+ 2084.0,
+ 869.0,
+ 2123.0,
+ 831.0,
+ 2123.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 447.0,
+ 1932.0,
+ 1250.0,
+ 1932.0,
+ 1250.0,
+ 1976.0,
+ 447.0,
+ 1976.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 225.0,
+ 661.0,
+ 225.0,
+ 661.0,
+ 268.0,
+ 294.0,
+ 268.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1378.0,
+ 1405.0,
+ 1378.0,
+ 1405.0,
+ 1416.0,
+ 293.0,
+ 1416.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1411.0,
+ 1405.0,
+ 1411.0,
+ 1405.0,
+ 1446.0,
+ 294.0,
+ 1446.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1440.0,
+ 1402.0,
+ 1440.0,
+ 1402.0,
+ 1473.0,
+ 295.0,
+ 1473.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1472.0,
+ 1405.0,
+ 1472.0,
+ 1405.0,
+ 1505.0,
+ 295.0,
+ 1505.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1503.0,
+ 955.0,
+ 1503.0,
+ 955.0,
+ 1537.0,
+ 293.0,
+ 1537.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 393.0,
+ 272.0,
+ 605.0,
+ 272.0,
+ 605.0,
+ 303.0,
+ 393.0,
+ 303.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 415.0,
+ 302.0,
+ 747.0,
+ 302.0,
+ 747.0,
+ 338.0,
+ 415.0,
+ 338.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 392.0,
+ 332.0,
+ 580.0,
+ 332.0,
+ 580.0,
+ 367.0,
+ 392.0,
+ 367.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 361.0,
+ 325.0,
+ 361.0,
+ 325.0,
+ 398.0,
+ 321.0,
+ 398.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 397.0,
+ 361.0,
+ 1113.0,
+ 361.0,
+ 1113.0,
+ 398.0,
+ 397.0,
+ 398.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 393.0,
+ 324.0,
+ 393.0,
+ 324.0,
+ 426.0,
+ 320.0,
+ 426.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 392.0,
+ 393.0,
+ 397.0,
+ 393.0,
+ 397.0,
+ 426.0,
+ 392.0,
+ 426.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 424.0,
+ 472.0,
+ 424.0,
+ 472.0,
+ 459.0,
+ 323.0,
+ 459.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 452.0,
+ 365.0,
+ 452.0,
+ 365.0,
+ 486.0,
+ 322.0,
+ 486.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 446.0,
+ 452.0,
+ 478.0,
+ 452.0,
+ 478.0,
+ 486.0,
+ 446.0,
+ 486.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 513.0,
+ 452.0,
+ 552.0,
+ 452.0,
+ 552.0,
+ 486.0,
+ 513.0,
+ 486.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 349.0,
+ 484.0,
+ 438.0,
+ 484.0,
+ 438.0,
+ 519.0,
+ 349.0,
+ 519.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 350.0,
+ 516.0,
+ 438.0,
+ 516.0,
+ 438.0,
+ 551.0,
+ 350.0,
+ 551.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 348.0,
+ 540.0,
+ 930.0,
+ 540.0,
+ 930.0,
+ 586.0,
+ 348.0,
+ 586.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 985.0,
+ 540.0,
+ 988.0,
+ 540.0,
+ 988.0,
+ 586.0,
+ 985.0,
+ 586.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 347.0,
+ 576.0,
+ 350.0,
+ 576.0,
+ 350.0,
+ 609.0,
+ 347.0,
+ 609.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 424.0,
+ 576.0,
+ 429.0,
+ 576.0,
+ 429.0,
+ 609.0,
+ 424.0,
+ 609.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 607.0,
+ 421.0,
+ 607.0,
+ 421.0,
+ 639.0,
+ 322.0,
+ 639.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 638.0,
+ 365.0,
+ 638.0,
+ 365.0,
+ 667.0,
+ 324.0,
+ 667.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 431.0,
+ 638.0,
+ 463.0,
+ 638.0,
+ 463.0,
+ 667.0,
+ 431.0,
+ 667.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 493.0,
+ 638.0,
+ 531.0,
+ 638.0,
+ 531.0,
+ 667.0,
+ 493.0,
+ 667.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 349.0,
+ 668.0,
+ 393.0,
+ 668.0,
+ 393.0,
+ 701.0,
+ 349.0,
+ 701.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 462.0,
+ 668.0,
+ 494.0,
+ 668.0,
+ 494.0,
+ 701.0,
+ 462.0,
+ 701.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 525.0,
+ 668.0,
+ 564.0,
+ 668.0,
+ 564.0,
+ 701.0,
+ 525.0,
+ 701.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 377.0,
+ 696.0,
+ 465.0,
+ 696.0,
+ 465.0,
+ 735.0,
+ 377.0,
+ 735.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 376.0,
+ 729.0,
+ 464.0,
+ 729.0,
+ 464.0,
+ 764.0,
+ 376.0,
+ 764.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 376.0,
+ 761.0,
+ 666.0,
+ 761.0,
+ 666.0,
+ 802.0,
+ 376.0,
+ 802.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 377.0,
+ 794.0,
+ 429.0,
+ 794.0,
+ 429.0,
+ 832.0,
+ 377.0,
+ 832.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 619.0,
+ 794.0,
+ 865.0,
+ 794.0,
+ 865.0,
+ 832.0,
+ 619.0,
+ 832.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 899.0,
+ 794.0,
+ 904.0,
+ 794.0,
+ 904.0,
+ 832.0,
+ 899.0,
+ 832.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 377.0,
+ 825.0,
+ 1048.0,
+ 825.0,
+ 1048.0,
+ 861.0,
+ 377.0,
+ 861.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 373.0,
+ 856.0,
+ 378.0,
+ 856.0,
+ 378.0,
+ 889.0,
+ 373.0,
+ 889.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 451.0,
+ 856.0,
+ 456.0,
+ 856.0,
+ 456.0,
+ 889.0,
+ 451.0,
+ 889.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 349.0,
+ 882.0,
+ 448.0,
+ 882.0,
+ 448.0,
+ 919.0,
+ 349.0,
+ 919.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 350.0,
+ 916.0,
+ 851.0,
+ 916.0,
+ 851.0,
+ 949.0,
+ 350.0,
+ 949.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 351.0,
+ 946.0,
+ 547.0,
+ 946.0,
+ 547.0,
+ 981.0,
+ 351.0,
+ 981.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 600.0,
+ 946.0,
+ 778.0,
+ 946.0,
+ 778.0,
+ 981.0,
+ 600.0,
+ 981.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 831.0,
+ 946.0,
+ 1052.0,
+ 946.0,
+ 1052.0,
+ 981.0,
+ 831.0,
+ 981.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 350.0,
+ 977.0,
+ 547.0,
+ 977.0,
+ 547.0,
+ 1013.0,
+ 350.0,
+ 1013.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 606.0,
+ 977.0,
+ 784.0,
+ 977.0,
+ 784.0,
+ 1013.0,
+ 606.0,
+ 1013.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 843.0,
+ 977.0,
+ 1070.0,
+ 977.0,
+ 1070.0,
+ 1013.0,
+ 843.0,
+ 1013.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 350.0,
+ 1004.0,
+ 709.0,
+ 1004.0,
+ 709.0,
+ 1044.0,
+ 350.0,
+ 1044.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 916.0,
+ 1004.0,
+ 919.0,
+ 1004.0,
+ 919.0,
+ 1044.0,
+ 916.0,
+ 1044.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 350.0,
+ 1037.0,
+ 930.0,
+ 1037.0,
+ 930.0,
+ 1074.0,
+ 350.0,
+ 1074.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 349.0,
+ 1071.0,
+ 393.0,
+ 1071.0,
+ 393.0,
+ 1103.0,
+ 349.0,
+ 1103.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 462.0,
+ 1071.0,
+ 494.0,
+ 1071.0,
+ 494.0,
+ 1103.0,
+ 462.0,
+ 1103.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 548.0,
+ 1071.0,
+ 586.0,
+ 1071.0,
+ 586.0,
+ 1103.0,
+ 548.0,
+ 1103.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 378.0,
+ 1101.0,
+ 483.0,
+ 1101.0,
+ 483.0,
+ 1134.0,
+ 378.0,
+ 1134.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 516.0,
+ 1101.0,
+ 771.0,
+ 1101.0,
+ 771.0,
+ 1134.0,
+ 516.0,
+ 1134.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 890.0,
+ 1101.0,
+ 1035.0,
+ 1101.0,
+ 1035.0,
+ 1134.0,
+ 890.0,
+ 1134.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 378.0,
+ 1131.0,
+ 483.0,
+ 1131.0,
+ 483.0,
+ 1165.0,
+ 378.0,
+ 1165.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 517.0,
+ 1131.0,
+ 772.0,
+ 1131.0,
+ 772.0,
+ 1165.0,
+ 517.0,
+ 1165.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 806.0,
+ 1131.0,
+ 950.0,
+ 1131.0,
+ 950.0,
+ 1165.0,
+ 806.0,
+ 1165.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 349.0,
+ 1162.0,
+ 448.0,
+ 1162.0,
+ 448.0,
+ 1198.0,
+ 349.0,
+ 1198.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1195.0,
+ 419.0,
+ 1195.0,
+ 419.0,
+ 1228.0,
+ 322.0,
+ 1228.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 447.0,
+ 1932.0,
+ 1250.0,
+ 1932.0,
+ 1250.0,
+ 1976.0,
+ 447.0,
+ 1976.0
+ ],
+ "score": 1.0,
+ "text": ""
+ }
+ ],
+ "page_info": {
+ "page_no": 13,
+ "width": 1700,
+ "height": 2200
+ }
+ },
+ {
+ "layout_dets": [
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1333,
+ 1404,
+ 1333,
+ 1404,
+ 1578,
+ 298,
+ 1578
+ ],
+ "score": 0.982
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1683,
+ 1404,
+ 1683,
+ 1404,
+ 1868,
+ 298,
+ 1868
+ ],
+ "score": 0.979
+ },
+ {
+ "category_id": 5,
+ "poly": [
+ 365,
+ 707,
+ 1334,
+ 707,
+ 1334,
+ 1211,
+ 365,
+ 1211
+ ],
+ "score": 0.973,
+ "html": "