Auto-sync: 2026-06-30 23:21:36 (part 2)

results/paper_analysis.json

@@ -1,6 +1,6 @@
 {
   "best_clean_key": "residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
-  "generated_utc": "2026-07-01T03:07:08+00:00",
+  "generated_utc": "2026-07-01T03:30:26+00:00",
   "mechanism_gap": {
     "best_clean_vs_direct_same_ckpt": 0.0776811594202898,
     "best_clean_vs_h16": 0.06318840579710144,
@@ -2209,9 +2209,101 @@
       "std_success": 0.012819330079707819
     },
     "residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_noopwgcontact_challenger001": {
+      "ci95_success": 0.020418303175285292,
       "label": "K4 compatible tangents, no-op bonus 0.03, near-miss global + no-op/wrong-gripper contact challenger",
-      "missing": true,
-      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmglobal_noopwgcontact_challenger0p01_summary.json"
+      "mean_action_mse_to_best": 0.4220015877709333,
+      "mean_progress": 0.561136360562357,
+      "mean_success": 0.34144927536231884,
+      "num_completed": 3,
+      "per_task_success": {
+        "LiftPegUpright-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.26340865087329823,
+          "std_success": 0.06057881385012102
+        },
+        "PickCube-v1": {
+          "mean_num_groups": 196.66666666666666,
+          "mean_success": 0.28808077992860603,
+          "std_success": 0.008588713738165605
+        },
+        "PullCube-v1": {
+          "mean_num_groups": 81.0,
+          "mean_success": 0.2059922533606744,
+          "std_success": 0.02531545694792789
+        },
+        "PushCube-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.7604066042026859,
+          "std_success": 0.08151850231627764
+        },
+        "StackCube-v1": {
+          "mean_num_groups": 93.33333333333333,
+          "mean_success": 0.19336959336959336,
+          "std_success": 0.08682910965604235
+        }
+      },
+      "seed_action_mse_to_best": {
+        "0": 0.4010052319470307,
+        "1": 0.4123674940139703,
+        "2": 0.45263203735179874
+      },
+      "seed_progress": {
+        "0": 0.5468972360149729,
+        "1": 0.5690277994780437,
+        "2": 0.5674840461940545
+      },
+      "seed_success": {
+        "0": 0.33217391304347826,
+        "1": 0.3443478260869565,
+        "2": 0.34782608695652173
+      },
+      "selected_candidate_type_counts": {
+        "retrieval_residual_policy_residual": 1208,
+        "retrieval_residual_residual_near_miss": 50,
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": 5,
+        "retrieval_residual_residual_no_op": 313,
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": 27,
+        "retrieval_residual_residual_wrong_gripper": 122
+      },
+      "selected_residual_scale_counts": {
+        "0.35": 1351,
+        "0.4": 70,
+        "0.45": 304
+      },
+      "selected_type_outcomes": {
+        "retrieval_residual_policy_residual": {
+          "count": 1208.0,
+          "mean_progress": 0.5502828951138272,
+          "success_rate": 0.33857615894039733
+        },
+        "retrieval_residual_residual_near_miss": {
+          "count": 50.0,
+          "mean_progress": 0.6465109135210514,
+          "success_rate": 0.46
+        },
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": {
+          "count": 5.0,
+          "mean_progress": 0.7557853102684021,
+          "success_rate": 0.6
+        },
+        "retrieval_residual_residual_no_op": {
+          "count": 313.0,
+          "mean_progress": 0.5516062982234592,
+          "success_rate": 0.26198083067092653
+        },
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": {
+          "count": 27.0,
+          "mean_progress": 0.6916872307106301,
+          "success_rate": 0.48148148148148145
+        },
+        "retrieval_residual_residual_wrong_gripper": {
+          "count": 122.0,
+          "mean_progress": 0.6211941464921168,
+          "success_rate": 0.48360655737704916
+        }
+      },
+      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmglobal_noopwgcontact_challenger0p01_summary.json",
+      "std_success": 0.008218809784787139
     },
     "residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_challenger001": {
       "ci95_success": 0.03920403094295293,
@@ -2505,9 +2597,101 @@
       "std_success": 0.015780472566743406
     },
     "residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpullstack_challenger001": {
+      "ci95_success": 0.03691512166355961,
       "label": "K4 compatible tangents, no-op bonus 0.03, near-miss global + wrong-gripper challenger on Pick/Pull/Stack",
-      "missing": true,
-      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmglobal_wgpickpullstack_challenger0p01_summary.json"
+      "mean_action_mse_to_best": 0.4207441219177259,
+      "mean_progress": 0.5743561674773712,
+      "mean_success": 0.36000000000000004,
+      "num_completed": 3,
+      "per_task_success": {
+        "LiftPegUpright-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.26340865087329823,
+          "std_success": 0.06057881385012102
+        },
+        "PickCube-v1": {
+          "mean_num_groups": 196.66666666666666,
+          "mean_success": 0.3305568224046485,
+          "std_success": 0.04196214617114575
+        },
+        "PullCube-v1": {
+          "mean_num_groups": 81.0,
+          "mean_success": 0.228271436166173,
+          "std_success": 0.03525140677829959
+        },
+        "PushCube-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.7604066042026859,
+          "std_success": 0.08151850231627764
+        },
+        "StackCube-v1": {
+          "mean_num_groups": 93.33333333333333,
+          "mean_success": 0.19336959336959336,
+          "std_success": 0.08682910965604235
+        }
+      },
+      "seed_action_mse_to_best": {
+        "0": 0.39735263349083455,
+        "1": 0.4139976060374275,
+        "2": 0.45088212622491564
+      },
+      "seed_progress": {
+        "0": 0.5613569916915689,
+        "1": 0.5718237993347904,
+        "2": 0.5898877114057541
+      },
+      "seed_success": {
+        "0": 0.3582608695652174,
+        "1": 0.34608695652173915,
+        "2": 0.37565217391304345
+      },
+      "selected_candidate_type_counts": {
+        "retrieval_residual_policy_residual": 1418,
+        "retrieval_residual_residual_near_miss": 91,
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": 5,
+        "retrieval_residual_residual_no_op": 38,
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": 27,
+        "retrieval_residual_residual_wrong_gripper": 146
+      },
+      "selected_residual_scale_counts": {
+        "0.35": 1492,
+        "0.4": 41,
+        "0.45": 192
+      },
+      "selected_type_outcomes": {
+        "retrieval_residual_policy_residual": {
+          "count": 1418.0,
+          "mean_progress": 0.5610577679915375,
+          "success_rate": 0.33709449929478136
+        },
+        "retrieval_residual_residual_near_miss": {
+          "count": 91.0,
+          "mean_progress": 0.5661385127926594,
+          "success_rate": 0.38461538461538464
+        },
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": {
+          "count": 5.0,
+          "mean_progress": 0.7557853102684021,
+          "success_rate": 0.6
+        },
+        "retrieval_residual_residual_no_op": {
+          "count": 38.0,
+          "mean_progress": 0.7757954887653652,
+          "success_rate": 0.5263157894736842
+        },
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": {
+          "count": 27.0,
+          "mean_progress": 0.6916872307106301,
+          "success_rate": 0.48148148148148145
+        },
+        "retrieval_residual_residual_wrong_gripper": {
+          "count": 146.0,
+          "mean_progress": 0.6282956086898642,
+          "success_rate": 0.4931506849315068
+        }
+      },
+      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmglobal_wgpickpullstack_challenger0p01_summary.json",
+      "std_success": 0.01485913694837829
     },
     "residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmwgchallenger001": {
       "ci95_success": 0.028111505659259583,
@@ -8256,6 +8440,200 @@
       "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4s045_safe_margin0p20_kernel_mean_by_type_noopbonus0p03_summary.json",
       "std_success": 0.010190374394925756
     },
+    "residual_k6_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001": {
+      "ci95_success": 0.029933913043478266,
+      "label": "K6 compatible tangents, no-op bonus 0.03, near-miss challenger gate 0.01",
+      "mean_action_mse_to_best": 0.4099807325078417,
+      "mean_progress": 0.5686452052022829,
+      "mean_success": 0.34956521739130436,
+      "num_completed": 3,
+      "per_task_success": {
+        "LiftPegUpright-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.26340865087329823,
+          "std_success": 0.06057881385012102
+        },
+        "PickCube-v1": {
+          "mean_num_groups": 196.66666666666666,
+          "mean_success": 0.31909606094388704,
+          "std_success": 0.03185195743735153
+        },
+        "PullCube-v1": {
+          "mean_num_groups": 81.0,
+          "mean_success": 0.19973038657249184,
+          "std_success": 0.03143110860617521
+        },
+        "PushCube-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.7538059441366793,
+          "std_success": 0.07965457487204111
+        },
+        "StackCube-v1": {
+          "mean_num_groups": 93.33333333333333,
+          "mean_success": 0.18697228697228696,
+          "std_success": 0.06568668859904422
+        }
+      },
+      "seed_action_mse_to_best": {
+        "0": 0.39932717701177234,
+        "1": 0.4013874253679229,
+        "2": 0.42922759514383
+      },
+      "seed_progress": {
+        "0": 0.5556146927902718,
+        "1": 0.5693354412492203,
+        "2": 0.5809854815673569
+      },
+      "seed_success": {
+        "0": 0.3426086956521739,
+        "1": 0.3426086956521739,
+        "2": 0.3634782608695652
+      },
+      "selected_candidate_type_counts": {
+        "retrieval_residual_policy_residual": 1524,
+        "retrieval_residual_residual_near_miss": 104,
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": 5,
+        "retrieval_residual_residual_no_op": 34,
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": 35,
+        "retrieval_residual_residual_wrong_gripper": 23
+      },
+      "selected_residual_scale_counts": {
+        "0.35": 1563,
+        "0.4": 13,
+        "0.45": 149
+      },
+      "selected_type_outcomes": {
+        "retrieval_residual_policy_residual": {
+          "count": 1524.0,
+          "mean_progress": 0.5664185173288562,
+          "success_rate": 0.3484251968503937
+        },
+        "retrieval_residual_residual_near_miss": {
+          "count": 104.0,
+          "mean_progress": 0.5384184337901667,
+          "success_rate": 0.3269230769230769
+        },
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": {
+          "count": 5.0,
+          "mean_progress": 0.9185634613037109,
+          "success_rate": 0.8
+        },
+        "retrieval_residual_residual_no_op": {
+          "count": 34.0,
+          "mean_progress": 0.6942972383078407,
+          "success_rate": 0.38235294117647056
+        },
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": {
+          "count": 35.0,
+          "mean_progress": 0.619267104778971,
+          "success_rate": 0.37142857142857144
+        },
+        "retrieval_residual_residual_wrong_gripper": {
+          "count": 23.0,
+          "mean_progress": 0.5140160597536875,
+          "success_rate": 0.34782608695652173
+        }
+      },
+      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k6_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json",
+      "std_success": 0.012049049096131323
+    },
+    "residual_k8_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001": {
+      "ci95_success": 0.028767893823112123,
+      "label": "K8 compatible tangents, no-op bonus 0.03, near-miss challenger gate 0.01",
+      "mean_action_mse_to_best": 0.4111984846284748,
+      "mean_progress": 0.5684753134327036,
+      "mean_success": 0.3501449275362319,
+      "num_completed": 3,
+      "per_task_success": {
+        "LiftPegUpright-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.26340865087329823,
+          "std_success": 0.06057881385012102
+        },
+        "PickCube-v1": {
+          "mean_num_groups": 196.66666666666666,
+          "mean_success": 0.3209885927277232,
+          "std_success": 0.028546221805186907
+        },
+        "PullCube-v1": {
+          "mean_num_groups": 81.0,
+          "mean_success": 0.19670894407736514,
+          "std_success": 0.01355096164672601
+        },
+        "PushCube-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.7505571718758995,
+          "std_success": 0.08230105593270941
+        },
+        "StackCube-v1": {
+          "mean_num_groups": 93.33333333333333,
+          "mean_success": 0.1937062937062937,
+          "std_success": 0.07732492963924059
+        }
+      },
+      "seed_action_mse_to_best": {
+        "0": 0.40068337346710586,
+        "1": 0.40238139059718536,
+        "2": 0.43053068982113313
+      },
+      "seed_progress": {
+        "0": 0.5535212510512413,
+        "1": 0.5703616362777741,
+        "2": 0.5815430529690955
+      },
+      "seed_success": {
+        "0": 0.3426086956521739,
+        "1": 0.3443478260869565,
+        "2": 0.3634782608695652
+      },
+      "selected_candidate_type_counts": {
+        "retrieval_residual_policy_residual": 1522,
+        "retrieval_residual_residual_near_miss": 109,
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": 5,
+        "retrieval_residual_residual_no_op": 30,
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": 38,
+        "retrieval_residual_residual_wrong_gripper": 21
+      },
+      "selected_residual_scale_counts": {
+        "0.35": 1558,
+        "0.4": 17,
+        "0.45": 150
+      },
+      "selected_type_outcomes": {
+        "retrieval_residual_policy_residual": {
+          "count": 1522.0,
+          "mean_progress": 0.566564130936272,
+          "success_rate": 0.34822601839684625
+        },
+        "retrieval_residual_residual_near_miss": {
+          "count": 109.0,
+          "mean_progress": 0.532924958487648,
+          "success_rate": 0.3211009174311927
+        },
+        "retrieval_residual_residual_near_miss+residual_wrong_gripper": {
+          "count": 5.0,
+          "mean_progress": 0.9184834241867066,
+          "success_rate": 0.8
+        },
+        "retrieval_residual_residual_no_op": {
+          "count": 30.0,
+          "mean_progress": 0.6870485415061315,
+          "success_rate": 0.4
+        },
+        "retrieval_residual_residual_no_op+residual_wrong_gripper": {
+          "count": 38.0,
+          "mean_progress": 0.5830513087934569,
+          "success_rate": 0.34210526315789475
+        },
+        "retrieval_residual_residual_wrong_gripper": {
+          "count": 21.0,
+          "mean_progress": 0.6124126100469203,
+          "success_rate": 0.47619047619047616
+        }
+      },
+      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k8_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json",
+      "std_success": 0.01157970107561633
+    },
     "residual_k8_fieldsoftmax_grid_noopbonus003": {
       "ci95_success": 0.033559967695396,
       "label": "K8 field-softmax tangent transport, scales 0.35/0.40/0.45, no-op bonus 0.03",

results/paper_analysis.md

@@ -1,6 +1,6 @@
 # Paper Analysis
 
-Generated: `2026-07-01T03:07:08+00:00`
+Generated: `2026-07-01T03:30:26+00:00`
 
 ## Main Seed Statistics
 
@@ -67,8 +67,10 @@ Generated: `2026-07-01T03:07:08+00:00`
 | residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + wrong-gripper challenger on Pick/Pull, gate 0.01 | 3 | 36.06% +/- 1.58 | +/- 3.92 | 57.41% | 0.419 | +6.32 pp |
 | residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_wgmargin003_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + wrong-gripper Pick/Pull margin 0.03 | 3 | 36.00% +/- 1.59 | +/- 3.96 | 57.39% | 0.417 | +6.26 pp |
 | residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_wgmargin005_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + wrong-gripper Pick/Pull margin 0.05 | 3 | 35.88% +/- 1.58 | +/- 3.92 | 57.31% | 0.414 | +6.14 pp |
-| residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpullstack_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + wrong-gripper challenger on Pick/Pull/Stack | 0 | missing | missing | missing | missing | missing |
-| residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_noopwgcontact_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + no-op/wrong-gripper contact challenger | 0 | missing | missing | missing | missing | missing |
+| residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpullstack_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + wrong-gripper challenger on Pick/Pull/Stack | 3 | 36.00% +/- 1.49 | +/- 3.69 | 57.44% | 0.421 | +6.26 pp |
+| residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_noopwgcontact_challenger001 | K4 compatible tangents, no-op bonus 0.03, near-miss global + no-op/wrong-gripper contact challenger | 3 | 34.14% +/- 0.82 | +/- 2.04 | 56.11% | 0.422 | +4.41 pp |
+| residual_k6_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001 | K6 compatible tangents, no-op bonus 0.03, near-miss challenger gate 0.01 | 3 | 34.96% +/- 1.20 | +/- 2.99 | 56.86% | 0.410 | +5.22 pp |
+| residual_k8_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001 | K8 compatible tangents, no-op bonus 0.03, near-miss challenger gate 0.01 | 3 | 35.01% +/- 1.16 | +/- 2.88 | 56.85% | 0.411 | +5.28 pp |
 | typed_proposal_lattice_types6_prepend_margin000 | Typed proposal lattice head, six families, policy-prepended margin 0.00 | 3 | 31.30% +/- 3.81 | +/- 9.48 | 54.03% | 0.480 | +1.57 pp |
 | typed_proposal_lattice_types6_prepend_margin005 | Typed proposal lattice head, six families, policy-prepended margin 0.05 | 3 | 32.17% +/- 3.05 | +/- 7.57 | 54.74% | 0.467 | +2.43 pp |
 | typed_proposal_lattice_types4safe_prepend_margin005 | Typed proposal lattice head, safe four families, policy-prepended margin 0.05 | 3 | 33.45% +/- 3.39 | +/- 8.43 | 55.47% | 0.456 | +3.71 pp |

results/paper_table_status.json

@@ -1686,14 +1686,14 @@
       "story_role": "oracle-trace contact-task calibration: wrong-gripper residuals are only challenger-eligible on Pick/Pull/Stack while near-miss stays global",
       "fallback_success": null,
       "pending_job": "14995028/14995047",
-      "path_exists": false,
-      "status": "pending",
-      "success": null,
-      "std_success": null,
+      "path_exists": true,
+      "status": "complete",
+      "success": 0.36,
+      "std_success": 0.01485913694837829,
       "completed_seeds": null,
-      "num_completed": null,
+      "num_completed": 3,
       "best_config": null,
-      "gain_vs_h16_policy": null
+      "gain_vs_h16_policy": 0.06260869565217392
     },
     {
       "key": "retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_noopwgcontact_challenger001",
@@ -1705,14 +1705,52 @@
       "story_role": "contact-task sparse-family challenger ablation using no-op and wrong-gripper residuals where oracle trace shows proposal headroom",
       "fallback_success": null,
       "pending_job": "14995045/14995048",
-      "path_exists": false,
-      "status": "pending",
-      "success": null,
-      "std_success": null,
+      "path_exists": true,
+      "status": "complete",
+      "success": 0.34144927536231884,
+      "std_success": 0.00821880978478714,
       "completed_seeds": null,
-      "num_completed": null,
+      "num_completed": 3,
       "best_config": null,
-      "gain_vs_h16_policy": null
+      "gain_vs_h16_policy": 0.04405797101449277
+    },
+    {
+      "key": "retrieval_residual_k6_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
+      "label": "K6 compatible residual retrieval, near-miss challenger gate 0.01",
+      "path": "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k6_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json",
+      "clean_deployment": "yes",
+      "same_state_proposals": "no",
+      "expert_proposal": "no",
+      "story_role": "neighborhood-size test for whether compatible transported tangent charts need more train-state support than K4",
+      "fallback_success": null,
+      "pending_job": "14995739/14995759",
+      "path_exists": true,
+      "status": "complete",
+      "success": 0.3495652173913043,
+      "std_success": 0.01204904909613132,
+      "completed_seeds": null,
+      "num_completed": 3,
+      "best_config": null,
+      "gain_vs_h16_policy": 0.05217391304347824
+    },
+    {
+      "key": "retrieval_residual_k8_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
+      "label": "K8 compatible residual retrieval, near-miss challenger gate 0.01",
+      "path": "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k8_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json",
+      "clean_deployment": "yes",
+      "same_state_proposals": "no",
+      "expert_proposal": "no",
+      "story_role": "larger-neighborhood transported tangent support test after K4 plateaued near 36%",
+      "fallback_success": null,
+      "pending_job": "14995741/14995761",
+      "path_exists": true,
+      "status": "complete",
+      "success": 0.3501449275362319,
+      "std_success": 0.01157970107561633,
+      "completed_seeds": null,
+      "num_completed": 3,
+      "best_config": null,
+      "gain_vs_h16_policy": 0.05275362318840582
     },
     {
       "key": "typed_proposal_lattice_types6_prepend_margin000",

results/paper_table_status.md

@@ -91,8 +91,10 @@ Baseline h=16 policy: 29.74%
 | retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_challenger001 | K4 compatible residual retrieval, near-miss global + wrong-gripper challenger on Pick/Pull, gate 0.01 | complete | 36.06% | +6.32 pp | yes | no | no | type-specific affordance gate: keep near-miss global while restricting wrong-gripper challenger to acquisition/contact tasks |
 | retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_wgmargin003_challenger001 | K4 compatible residual retrieval, near-miss global + wrong-gripper Pick/Pull margin 0.03 | complete | 36.00% | +6.26 pp | yes | no | no | type-specific affordance margin: keep near-miss globally sensitive while requiring stronger wrong-gripper evidence only on Pick/Pull |
 | retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpull_wgmargin005_challenger001 | K4 compatible residual retrieval, near-miss global + wrong-gripper Pick/Pull margin 0.05 | complete | 35.88% | +6.14 pp | yes | no | no | upper setting for type-specific wrong-gripper evidence on Pick/Pull while preserving global near-miss calibration |
-| retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpullstack_challenger001 | K4 compatible residual retrieval, near-miss global + wrong-gripper challenger on Pick/Pull/Stack | pending 14995028/14995047 | pending | pending | yes | no | no | oracle-trace contact-task calibration: wrong-gripper residuals are only challenger-eligible on Pick/Pull/Stack while near-miss stays global |
-| retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_noopwgcontact_challenger001 | K4 compatible residual retrieval, near-miss global + no-op/wrong-gripper contact challenger | pending 14995045/14995048 | pending | pending | yes | no | no | contact-task sparse-family challenger ablation using no-op and wrong-gripper residuals where oracle trace shows proposal headroom |
+| retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_wgpickpullstack_challenger001 | K4 compatible residual retrieval, near-miss global + wrong-gripper challenger on Pick/Pull/Stack | complete | 36.00% | +6.26 pp | yes | no | no | oracle-trace contact-task calibration: wrong-gripper residuals are only challenger-eligible on Pick/Pull/Stack while near-miss stays global |
+| retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmglobal_noopwgcontact_challenger001 | K4 compatible residual retrieval, near-miss global + no-op/wrong-gripper contact challenger | complete | 34.14% | +4.41 pp | yes | no | no | contact-task sparse-family challenger ablation using no-op and wrong-gripper residuals where oracle trace shows proposal headroom |
+| retrieval_residual_k6_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001 | K6 compatible residual retrieval, near-miss challenger gate 0.01 | complete | 34.96% | +5.22 pp | yes | no | no | neighborhood-size test for whether compatible transported tangent charts need more train-state support than K4 |
+| retrieval_residual_k8_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001 | K8 compatible residual retrieval, near-miss challenger gate 0.01 | complete | 35.01% | +5.28 pp | yes | no | no | larger-neighborhood transported tangent support test after K4 plateaued near 36% |
 | typed_proposal_lattice_types6_prepend_margin000 | Typed proposal lattice head, six families, policy-prepended margin 0.00 | complete | 31.30% | +1.57 pp | yes | no | no | model-generated counterfactual proposal support test; field scores typed proposals rather than retrieved train-state residuals |
 | typed_proposal_lattice_types6_prepend_margin005 | Typed proposal lattice head, six families, policy-prepended margin 0.05 | complete | 32.17% | +2.43 pp | yes | no | no | model-generated counterfactual proposal support test with a conservative policy-abstention margin |
 | typed_proposal_lattice_types4safe_prepend_margin005 | Typed proposal lattice head, safe four families, policy-prepended margin 0.05 | complete | 33.45% | +3.71 pp | yes | no | no | anti-goal proposal-family mask test for the model-generated typed lattice |

scripts/build_paper_analysis.py

@@ -529,6 +529,22 @@ METHODS = [
             "k4_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmglobal_noopwgcontact_challenger0p01_summary.json"
         ),
     ),
+    MethodSpec(
+        key="residual_k6_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
+        label="K6 compatible tangents, no-op bonus 0.03, near-miss challenger gate 0.01",
+        summary_path=(
+            "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
+            "k6_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json"
+        ),
+    ),
+    MethodSpec(
+        key="residual_k8_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
+        label="K8 compatible tangents, no-op bonus 0.03, near-miss challenger gate 0.01",
+        summary_path=(
+            "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
+            "k8_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json"
+        ),
+    ),
     MethodSpec(
         key="typed_proposal_lattice_types6_prepend_margin000",
         label="Typed proposal lattice head, six families, policy-prepended margin 0.00",

scripts/build_paper_table_status.py

@@ -915,6 +915,26 @@ SPECS = [
         story_role="contact-task sparse-family challenger ablation using no-op and wrong-gripper residuals where oracle trace shows proposal headroom",
         pending_job="14995045/14995048",
     ),
+    ResultSpec(
+        key="retrieval_residual_k6_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
+        label="K6 compatible residual retrieval, near-miss challenger gate 0.01",
+        path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k6_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json",
+        clean_deployment="yes",
+        same_state_proposals="no",
+        expert_proposal="no",
+        story_role="neighborhood-size test for whether compatible transported tangent charts need more train-state support than K4",
+        pending_job="14995739/14995759",
+    ),
+    ResultSpec(
+        key="retrieval_residual_k8_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
+        label="K8 compatible residual retrieval, near-miss challenger gate 0.01",
+        path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k8_composemasked_dropnmnoop_grid035040045_safe_margin0p20_noopbonus0p03_nmchallenger0p01_summary.json",
+        clean_deployment="yes",
+        same_state_proposals="no",
+        expert_proposal="no",
+        story_role="larger-neighborhood transported tangent support test after K4 plateaued near 36%",
+        pending_job="14995741/14995761",
+    ),
     ResultSpec(
         key="typed_proposal_lattice_types6_prepend_margin000",
         label="Typed proposal lattice head, six families, policy-prepended margin 0.00",