| { |
| "status": "passed", |
| "validation_scope": "Analysis-only reproduction from a clean local copy of the staged archive. Original numerical statistics were compared at rtol=1e-8, atol=1e-10. Model outputs and original research records were not regenerated or modified. Public-download verification is a separate check.", |
| "checks": [ |
| { |
| "name": "controlled_readout", |
| "command": "python reproduction/reproduce_controlled.py", |
| "exit_code": 0, |
| "seconds": 31.84, |
| "numeric_values_compared": 58740, |
| "differences": [], |
| "status": "passed", |
| "log": "controlled_readout.log" |
| }, |
| { |
| "name": "target_free_classification", |
| "command": "python scripts/analyze_candidate_nonsteering.py", |
| "exit_code": 0, |
| "seconds": 20.72, |
| "numeric_values_compared": 331, |
| "differences": [], |
| "status": "passed", |
| "log": "target_free_classification.log" |
| }, |
| { |
| "name": "broad_steering", |
| "command": "python scripts/analyze_semantic_steering_v3.py", |
| "exit_code": 0, |
| "seconds": 3.89, |
| "numeric_values_compared": 260, |
| "differences": [], |
| "status": "passed", |
| "log": "broad_steering.log" |
| }, |
| { |
| "name": "grain_steering", |
| "command": "python scripts/analyze_relational_grain_steering_v4.py", |
| "exit_code": 0, |
| "seconds": 1.84, |
| "numeric_values_compared": 47, |
| "differences": [], |
| "status": "passed", |
| "log": "grain_steering.log" |
| }, |
| { |
| "name": "full_state_patching", |
| "command": "python scripts/analyze_counterfactual_activation_patching.py", |
| "exit_code": 0, |
| "seconds": 1.8, |
| "numeric_values_compared": 121, |
| "differences": [], |
| "status": "passed", |
| "log": "full_state_patching.log" |
| }, |
| { |
| "name": "cross_mechanism_patching", |
| "command": "python scripts/analyze_cross_mechanism_activation_patching.py", |
| "exit_code": 0, |
| "seconds": 1.28, |
| "numeric_values_compared": 144, |
| "differences": [], |
| "status": "passed", |
| "log": "cross_mechanism_patching.log" |
| }, |
| { |
| "name": "supplied_equation_60_laws", |
| "command": "python scripts/analyze_neutral_anchored_relational_benchmark.py", |
| "exit_code": 0, |
| "seconds": 33.97, |
| "numeric_values_compared": 35, |
| "differences": [], |
| "status": "passed", |
| "log": "supplied_equation_60_laws.log" |
| }, |
| { |
| "name": "equation_free_4b", |
| "command": "python scripts/analyze_equation_free_relational_benchmark_amended.py", |
| "exit_code": 0, |
| "seconds": 33.69, |
| "numeric_values_compared": 162, |
| "differences": [], |
| "status": "passed", |
| "log": "equation_free_4b.log" |
| }, |
| { |
| "name": "equation_free_scale_comparison", |
| "command": "python scripts/analyze_gemma_equation_free_multiscale.py", |
| "exit_code": 0, |
| "seconds": 3.49, |
| "numeric_values_compared": 92, |
| "differences": [], |
| "status": "passed", |
| "log": "equation_free_scale_comparison.log" |
| }, |
| { |
| "name": "readability_scale_comparison", |
| "command": "python scripts/analyze_gemma_multiscale_heldout_v1.py", |
| "exit_code": 0, |
| "seconds": 44.08, |
| "numeric_values_compared": 265, |
| "differences": [], |
| "status": "passed", |
| "log": "readability_scale_comparison.log" |
| }, |
| { |
| "name": "coordinate_interventions", |
| "command": "python scripts/analyze_causal_relational_knockout.py", |
| "exit_code": 0, |
| "seconds": 2.68, |
| "numeric_values_compared": 488, |
| "differences": [], |
| "status": "passed", |
| "log": "coordinate_interventions.log" |
| }, |
| { |
| "name": "coordinate_integrity", |
| "command": "python scripts/validate_causal_relational_knockout.py", |
| "exit_code": 0, |
| "seconds": 0.18, |
| "numeric_values_compared": 0, |
| "differences": [], |
| "status": "passed", |
| "log": "coordinate_integrity.log" |
| }, |
| { |
| "name": "lexical_adversarial", |
| "command": "python scripts/analyze_lexical_adversarial_representation.py", |
| "exit_code": 0, |
| "seconds": 17.15, |
| "numeric_values_compared": 132, |
| "differences": [], |
| "status": "passed", |
| "log": "lexical_adversarial.log" |
| }, |
| { |
| "name": "late_replication", |
| "command": "python scripts/analyze_late_physics_replication.py", |
| "exit_code": 0, |
| "seconds": 29.52, |
| "numeric_values_compared": 203, |
| "differences": [], |
| "status": "passed", |
| "log": "late_replication.log" |
| }, |
| { |
| "name": "answer_scaffold", |
| "command": "python scripts/analyze_answer_scaffold_audit.py", |
| "exit_code": 0, |
| "seconds": 5.41, |
| "numeric_values_compared": 32, |
| "differences": [], |
| "status": "passed", |
| "log": "answer_scaffold.log" |
| }, |
| { |
| "name": "answer_code", |
| "command": "python scripts/analyze_answer_code_binding.py", |
| "exit_code": 0, |
| "seconds": 35.38, |
| "numeric_values_compared": 240, |
| "differences": [], |
| "status": "passed", |
| "log": "answer_code.log" |
| }, |
| { |
| "name": "graph_integrity", |
| "command": "python scripts/audit_graph_topology_rigorous.py", |
| "exit_code": 0, |
| "seconds": 0.35, |
| "numeric_values_compared": 0, |
| "differences": [], |
| "status": "passed", |
| "log": "graph_integrity.log" |
| }, |
| { |
| "name": "graph_generalization_integrity", |
| "command": "python scripts/validate_graph_generalization_audit.py", |
| "exit_code": 0, |
| "seconds": 0.61, |
| "numeric_values_compared": 0, |
| "differences": [], |
| "status": "passed", |
| "log": "graph_generalization_integrity.log" |
| }, |
| { |
| "name": "graph_figures", |
| "command": "python scripts/plot_graph_generalization_audit.py", |
| "exit_code": 0, |
| "seconds": 4.64, |
| "numeric_values_compared": 0, |
| "differences": [], |
| "status": "passed", |
| "log": "graph_figures.log" |
| }, |
| { |
| "name": "latent_geometry", |
| "command": "python scripts/analyze_materials_latent_geometry.py", |
| "exit_code": 0, |
| "seconds": 119.02, |
| "numeric_values_compared": 304, |
| "differences": [], |
| "status": "passed", |
| "log": "latent_geometry.log" |
| }, |
| { |
| "name": "equation_free_12b", |
| "command": "python reproduction/reproduce_scale.py --root <working-copy> --scale 12b", |
| "exit_code": 0, |
| "seconds": 34.33, |
| "numeric_values_compared": 47, |
| "differences": [], |
| "status": "passed", |
| "log": "equation_free_12b.log" |
| }, |
| { |
| "name": "equation_free_31b", |
| "command": "python reproduction/reproduce_scale.py --root <working-copy> --scale 31b", |
| "exit_code": 0, |
| "seconds": 34.27, |
| "numeric_values_compared": 47, |
| "differences": [], |
| "status": "passed", |
| "log": "equation_free_31b.log" |
| } |
| ], |
| "numeric_values_compared": 61690, |
| "limitations": [ |
| "The saved graph-network training outputs were integrity-checked and used to rebuild publication figures; graph-network training was not repeated.", |
| "The original Gemma inference, lens fitting, and external automated-rater calls were not repeated. The automated-rater analysis was recomputed from cached responses.", |
| "Original protocol terminology and validator labels are historical records. The SI and CHRONOLOGY documentation provide the corrected chronology and overlap interpretation." |
| ] |
| } |
|
|
