results: record mixed-traffic short failure

1 files changed, 64 insertions, 0 deletions

diff --git a/results/kp_innovation_short_gate.json b/results/kp_innovation_short_gate.json
new file mode 100644
index 0000000..87cb02c
--- /dev/null
+++ b/results/kp_innovation_short_gate.json
@@ -0,0 +1,64 @@
+{
+  "checks": {
+    "each_affine_mac_count_at_most_1.40x_bp": true,
+    "elementwise_cost_reported": true,
+    "epoch11_to20_feedback_cosine_at_least_0.70": false,
+    "final_feedback_cosine_at_least_0.80": false,
+    "initial_layer_ratio_error_at_most_1e-5": true,
+    "innovation_accuracy_at_least_0.75": false,
+    "innovation_early_alignment_at_least_0.70": false,
+    "innovation_gain_over_matched_at_least_0.03": false,
+    "innovation_gain_over_raw_at_least_0.05": false,
+    "innovation_within_5_points_of_clean_kp": false,
+    "matched_norm_relative_error_at_most_1e-6": false,
+    "post_warmup_predictor_residual_ratio_at_most_0.25": true,
+    "records_trajectories_and_diagnostics_finite": false,
+    "same_network_alignment_gain_over_raw_at_least_0.15": false,
+    "zero_task_loss_queries": true
+  },
+  "confirmation_test_seeds_touched": false,
+  "full_validation_opened": false,
+  "metrics": {
+    "accuracy": {
+      "innovation": 0.1,
+      "matched": 0.1,
+      "raw": 0.1
+    },
+    "clean_kp_accuracy": 0.8266,
+    "elementwise_operations_estimate": {
+      "innovation": 1283917873152,
+      "matched": 2131789873152,
+      "raw": 1283917873152
+    },
+    "epoch11_to20_feedback_forward_cosine": NaN,
+    "final_feedback_forward_cosine": NaN,
+    "innovation_early_third_alignment": NaN,
+    "innovation_gain_over_matched": 0.0,
+    "innovation_gain_over_raw": 0.0,
+    "logical_batch_loss_queries": {
+      "innovation": 0,
+      "matched": 0,
+      "raw": 0
+    },
+    "mac_ratio_to_bp": {
+      "innovation": 1.327079721520482,
+      "matched": 1.327079721520482,
+      "raw": 1.327079721520482
+    },
+    "matched_norm_relative_error": NaN,
+    "max_initial_layer_ratio_error": 4.76837158203125e-07,
+    "max_post_warmup_predictor_residual_rms_ratio": 0.13994200354453204,
+    "same_innovation_network_raw_early_third_alignment": NaN,
+    "source_commit": "412314e3ed7bca666a41876980b479456f5e733d",
+    "total_macs": {
+      "innovation": 145299049750528,
+      "matched": 145299049750528,
+      "raw": 145299049750528
+    }
+  },
+  "protocol": "kp_mixed_traffic_short_v1",
+  "review_score_after": 5,
+  "review_score_before": 5,
+  "score_change_rule": "a one-seed short mechanism screen cannot raise the review score",
+  "status": "failed"
+}