anhtld commited on
Commit
049d149
·
verified ·
1 Parent(s): dbcbddf

Auto-sync: 2026-06-29 06:43:37

Browse files
scripts/build_paper_analysis.py CHANGED
@@ -211,6 +211,46 @@ METHODS = [
211
  "k4_grid035040045_safe_margin0p20_mean_by_type_srcscorebonus0p02_summary.json"
212
  ),
213
  ),
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
214
  MethodSpec(
215
  key="residual_k4_consensus_grid035040045_noopbonus003_l2penalty005",
216
  label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, no-op bonus 0.03, action L2 penalty 0.05",
 
211
  "k4_grid035040045_safe_margin0p20_mean_by_type_srcscorebonus0p02_summary.json"
212
  ),
213
  ),
214
+ MethodSpec(
215
+ key="residual_k4_consensus_grid035040045_srcadvbonus002",
216
+ label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, source-advantage bonus 0.02",
217
+ summary_path=(
218
+ "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
219
+ "k4_grid035040045_safe_margin0p20_mean_by_type_srcadvbonus0p02_summary.json"
220
+ ),
221
+ ),
222
+ MethodSpec(
223
+ key="residual_k4_consensus_grid035040045_srcadvbonus005",
224
+ label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, source-advantage bonus 0.05",
225
+ summary_path=(
226
+ "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
227
+ "k4_grid035040045_safe_margin0p20_mean_by_type_srcadvbonus0p05_summary.json"
228
+ ),
229
+ ),
230
+ MethodSpec(
231
+ key="residual_k4_consensus_grid035040045_noopbonus003_srcadvbonus002",
232
+ label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, no-op bonus 0.03, source-advantage bonus 0.02",
233
+ summary_path=(
234
+ "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
235
+ "k4_grid035040045_safe_margin0p20_mean_by_type_noopbonus0p03_srcadvbonus0p02_summary.json"
236
+ ),
237
+ ),
238
+ MethodSpec(
239
+ key="residual_k4_consensus_grid035040045_srcadvgate000",
240
+ label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, source-advantage gate >= 0.0",
241
+ summary_path=(
242
+ "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
243
+ "k4_grid035040045_safe_margin0p20_mean_by_type_srcadvgate0p0_summary.json"
244
+ ),
245
+ ),
246
+ MethodSpec(
247
+ key="residual_k4_consensus_grid035040045_noopbonus003_srcadvgate000",
248
+ label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, no-op bonus 0.03, source-advantage gate >= 0.0",
249
+ summary_path=(
250
+ "h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_"
251
+ "k4_grid035040045_safe_margin0p20_mean_by_type_noopbonus0p03_srcadvgate0p0_summary.json"
252
+ ),
253
+ ),
254
  MethodSpec(
255
  key="residual_k4_consensus_grid035040045_noopbonus003_l2penalty005",
256
  label="K4 mean-by-type tangent consensus, scales 0.35/0.40/0.45, no-op bonus 0.03, action L2 penalty 0.05",
scripts/build_paper_table_status.py CHANGED
@@ -525,6 +525,56 @@ SPECS = [
525
  story_role="scale-grid diagnostic for measured-prior mean-consensus residual transport",
526
  pending_job="14897989/14897991",
527
  ),
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
528
  ResultSpec(
529
  key="retrieval_residual_k4_mean_grid035040045_noopbonus003_l2penalty005",
530
  label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, no-op bonus 0.03, action L2 penalty 0.05",
 
525
  story_role="scale-grid diagnostic for measured-prior mean-consensus residual transport",
526
  pending_job="14897989/14897991",
527
  ),
528
+ ResultSpec(
529
+ key="retrieval_residual_k4_mean_grid035040045_srcadvbonus002",
530
+ label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, source-advantage bonus 0.02",
531
+ path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_grid035040045_safe_margin0p20_mean_by_type_srcadvbonus0p02_summary.json",
532
+ clean_deployment="yes",
533
+ same_state_proposals="no",
534
+ expert_proposal="no",
535
+ story_role="source-local utility-lift prior for mean-consensus residual transport",
536
+ pending_job="14902706/14902707",
537
+ ),
538
+ ResultSpec(
539
+ key="retrieval_residual_k4_mean_grid035040045_srcadvbonus005",
540
+ label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, source-advantage bonus 0.05",
541
+ path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_grid035040045_safe_margin0p20_mean_by_type_srcadvbonus0p05_summary.json",
542
+ clean_deployment="yes",
543
+ same_state_proposals="no",
544
+ expert_proposal="no",
545
+ story_role="source-local utility-lift prior for mean-consensus residual transport",
546
+ pending_job="14902709/14902711",
547
+ ),
548
+ ResultSpec(
549
+ key="retrieval_residual_k4_mean_grid035040045_noopbonus003_srcadvbonus002",
550
+ label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, no-op bonus 0.03, source-advantage bonus 0.02",
551
+ path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_grid035040045_safe_margin0p20_mean_by_type_noopbonus0p03_srcadvbonus0p02_summary.json",
552
+ clean_deployment="yes",
553
+ same_state_proposals="no",
554
+ expert_proposal="no",
555
+ story_role="source-local utility-lift calibration on the current best typed prior",
556
+ pending_job="14902713/14902715",
557
+ ),
558
+ ResultSpec(
559
+ key="retrieval_residual_k4_mean_grid035040045_srcadvgate000",
560
+ label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, source-advantage gate >= 0.0",
561
+ path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_grid035040045_safe_margin0p20_mean_by_type_srcadvgate0p0_summary.json",
562
+ clean_deployment="yes",
563
+ same_state_proposals="no",
564
+ expert_proposal="no",
565
+ story_role="source-local utility-lift gate for mean-consensus residual transport",
566
+ pending_job="14902717/14902719",
567
+ ),
568
+ ResultSpec(
569
+ key="retrieval_residual_k4_mean_grid035040045_noopbonus003_srcadvgate000",
570
+ label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, no-op bonus 0.03, source-advantage gate >= 0.0",
571
+ path="h16_policy_ckpt_near_miss_policy_bc5_bestpt_retrieval_residual_k4_grid035040045_safe_margin0p20_mean_by_type_noopbonus0p03_srcadvgate0p0_summary.json",
572
+ clean_deployment="yes",
573
+ same_state_proposals="no",
574
+ expert_proposal="no",
575
+ story_role="source-local utility-lift gate on the current best typed prior",
576
+ pending_job="14902721/14902723",
577
+ ),
578
  ResultSpec(
579
  key="retrieval_residual_k4_mean_grid035040045_noopbonus003_l2penalty005",
580
  label="K4 mean-by-type residual retrieval, scales 0.35/0.40/0.45, margin 0.20, no-op bonus 0.03, action L2 penalty 0.05",