From 0d167497b75031fae67b71fa1e7ab12e57c761bb Mon Sep 17 00:00:00 2001
From: YurenHao0426 <Blackhao0426@gmail.com>
Date: Mon, 10 Aug 2026 10:50:16 -0500
Subject: results: close BabyAI population predictor rescue

---
 results/babyai_shared/population_p3/seed4101.json | 515 ++++++++++++++++++++++
 1 file changed, 515 insertions(+)
 create mode 100644 results/babyai_shared/population_p3/seed4101.json

(limited to 'results/babyai_shared/population_p3/seed4101.json')

diff --git a/results/babyai_shared/population_p3/seed4101.json b/results/babyai_shared/population_p3/seed4101.json
new file mode 100644
index 0000000..75a0b0c
--- /dev/null
+++ b/results/babyai_shared/population_p3/seed4101.json
@@ -0,0 +1,515 @@
+{
+  "audits": [
+    {
+      "epoch": 0,
+      "layers": [
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.20984283089637756,
+            "mean_per_cell_r2": 0.9767748713493347,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.09602414071559906
+          },
+          "holdout": {
+            "context_rms": 0.20962461829185486,
+            "mean_per_cell_r2": 0.9725972414016724,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1049511507153511
+          },
+          "layer": 0,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.2008315920829773,
+            "mean_per_cell_r2": 0.9949333667755127,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.04755431041121483
+          },
+          "holdout": {
+            "context_rms": 0.20081116259098053,
+            "mean_per_cell_r2": 0.9941949844360352,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.05143459141254425
+          },
+          "layer": 1,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19726455211639404,
+            "mean_per_cell_r2": 0.997251570224762,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.03510858491063118
+          },
+          "holdout": {
+            "context_rms": 0.19765536487102509,
+            "mean_per_cell_r2": 0.996864914894104,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.03769101947546005
+          },
+          "layer": 2,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19451536238193512,
+            "mean_per_cell_r2": 0.9981643557548523,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.029410211369395256
+          },
+          "holdout": {
+            "context_rms": 0.194533109664917,
+            "mean_per_cell_r2": 0.997877836227417,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.031882479786872864
+          },
+          "layer": 3,
+          "teaching_observations": 0
+        }
+      ]
+    },
+    {
+      "epoch": 1,
+      "layers": [
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.20984283089637756,
+            "mean_per_cell_r2": 0.8648810386657715,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.23362895846366882
+          },
+          "holdout": {
+            "context_rms": 0.20962461829185486,
+            "mean_per_cell_r2": 0.8518301844596863,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.24587911367416382
+          },
+          "layer": 0,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.2008315920829773,
+            "mean_per_cell_r2": 0.9841650128364563,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.08432316780090332
+          },
+          "holdout": {
+            "context_rms": 0.20081116259098053,
+            "mean_per_cell_r2": 0.9819508790969849,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.09081554412841797
+          },
+          "layer": 1,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19726455211639404,
+            "mean_per_cell_r2": 0.9874575138092041,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.07510098814964294
+          },
+          "holdout": {
+            "context_rms": 0.19765536487102509,
+            "mean_per_cell_r2": 0.9855629205703735,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.08102375268936157
+          },
+          "layer": 2,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19451536238193512,
+            "mean_per_cell_r2": 0.9913246631622314,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.06404691189527512
+          },
+          "holdout": {
+            "context_rms": 0.194533109664917,
+            "mean_per_cell_r2": 0.9899281859397888,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.06954371184110641
+          },
+          "layer": 3,
+          "teaching_observations": 0
+        }
+      ]
+    },
+    {
+      "epoch": 5,
+      "layers": [
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.20984283089637756,
+            "mean_per_cell_r2": 0.8257500529289246,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.2655547857284546
+          },
+          "holdout": {
+            "context_rms": 0.20962461829185486,
+            "mean_per_cell_r2": 0.800991415977478,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.2851858139038086
+          },
+          "layer": 0,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.2008315920829773,
+            "mean_per_cell_r2": 0.9562175273895264,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.14005239307880402
+          },
+          "holdout": {
+            "context_rms": 0.20081116259098053,
+            "mean_per_cell_r2": 0.9500809907913208,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.15084101259708405
+          },
+          "layer": 1,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19726455211639404,
+            "mean_per_cell_r2": 0.9649862051010132,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.12556372582912445
+          },
+          "holdout": {
+            "context_rms": 0.19765536487102509,
+            "mean_per_cell_r2": 0.9601048827171326,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.13468508422374725
+          },
+          "layer": 2,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19451536238193512,
+            "mean_per_cell_r2": 0.9821997880935669,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.09148777276277542
+          },
+          "holdout": {
+            "context_rms": 0.194533109664917,
+            "mean_per_cell_r2": 0.9794628024101257,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.09917064756155014
+          },
+          "layer": 3,
+          "teaching_observations": 0
+        }
+      ]
+    },
+    {
+      "epoch": 10,
+      "layers": [
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.20984283089637756,
+            "mean_per_cell_r2": 0.8274786472320557,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.26442500948905945
+          },
+          "holdout": {
+            "context_rms": 0.20962461829185486,
+            "mean_per_cell_r2": 0.8023961782455444,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.28441107273101807
+          },
+          "layer": 0,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.2008315920829773,
+            "mean_per_cell_r2": 0.9351796507835388,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.17075972259044647
+          },
+          "holdout": {
+            "context_rms": 0.20081116259098053,
+            "mean_per_cell_r2": 0.9255064725875854,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1848365217447281
+          },
+          "layer": 1,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19726455211639404,
+            "mean_per_cell_r2": 0.9514586925506592,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1478380709886551
+          },
+          "holdout": {
+            "context_rms": 0.19765536487102509,
+            "mean_per_cell_r2": 0.943648099899292,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1601058840751648
+          },
+          "layer": 2,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19451536238193512,
+            "mean_per_cell_r2": 0.9786931872367859,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.10017189383506775
+          },
+          "holdout": {
+            "context_rms": 0.194533109664917,
+            "mean_per_cell_r2": 0.9755187630653381,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.10835007578134537
+          },
+          "layer": 3,
+          "teaching_observations": 0
+        }
+      ]
+    },
+    {
+      "epoch": 20,
+      "layers": [
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.20984283089637756,
+            "mean_per_cell_r2": 0.8492339253425598,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.24697861075401306
+          },
+          "holdout": {
+            "context_rms": 0.20962461829185486,
+            "mean_per_cell_r2": 0.826335072517395,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.2665298283100128
+          },
+          "layer": 0,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.2008315920829773,
+            "mean_per_cell_r2": 0.9172567129135132,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1933620125055313
+          },
+          "holdout": {
+            "context_rms": 0.20081116259098053,
+            "mean_per_cell_r2": 0.9072909951210022,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.206399068236351
+          },
+          "layer": 1,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19726455211639404,
+            "mean_per_cell_r2": 0.9384824633598328,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1660328060388565
+          },
+          "holdout": {
+            "context_rms": 0.19765536487102509,
+            "mean_per_cell_r2": 0.9301103353500366,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.1780754029750824
+          },
+          "layer": 2,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19451536238193512,
+            "mean_per_cell_r2": 0.9720658659934998,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.11473160237073898
+          },
+          "holdout": {
+            "context_rms": 0.194533109664917,
+            "mean_per_cell_r2": 0.9683250784873962,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.12316417694091797
+          },
+          "layer": 3,
+          "teaching_observations": 0
+        }
+      ]
+    },
+    {
+      "epoch": 40,
+      "layers": [
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.20984283089637756,
+            "mean_per_cell_r2": 0.7983269691467285,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.28594422340393066
+          },
+          "holdout": {
+            "context_rms": 0.20962461829185486,
+            "mean_per_cell_r2": 0.7654898762702942,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.310098260641098
+          },
+          "layer": 0,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.2008315920829773,
+            "mean_per_cell_r2": 0.8684720993041992,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.24367430806159973
+          },
+          "holdout": {
+            "context_rms": 0.20081116259098053,
+            "mean_per_cell_r2": 0.851003885269165,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.26162275671958923
+          },
+          "layer": 1,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19726455211639404,
+            "mean_per_cell_r2": 0.9062105417251587,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.20594285428524017
+          },
+          "holdout": {
+            "context_rms": 0.19765536487102509,
+            "mean_per_cell_r2": 0.8948484659194946,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.2192971259355545
+          },
+          "layer": 2,
+          "teaching_observations": 0
+        },
+        {
+          "action_observations": 0,
+          "calibration": {
+            "context_rms": 0.19451536238193512,
+            "mean_per_cell_r2": 0.9705915451049805,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.11770590394735336
+          },
+          "holdout": {
+            "context_rms": 0.194533109664917,
+            "mean_per_cell_r2": 0.9667372107505798,
+            "observations": 4096,
+            "residual_context_rms_ratio": 0.12624777853488922
+          },
+          "layer": 3,
+          "teaching_observations": 0
+        }
+      ]
+    }
+  ],
+  "calibration_examples": 4096,
+  "checks": {
+    "holdout_r2_at_least_0p8_every_epoch_and_layer": false,
+    "holdout_residual_ratio_at_most_0p25_every_epoch_and_layer": false,
+    "zero_action_or_teaching_observations": true
+  },
+  "config": {
+    "action_dim": 7,
+    "context_gain": 1.0,
+    "hidden_layers": 4,
+    "input_dim": 3964,
+    "learning_rate": 0.03,
+    "mission_dim": 23,
+    "momentum": 0.9,
+    "reciprocal_learning_rate": 0.03,
+    "weight_decay": 0.0001,
+    "width": 256
+  },
+  "epoch_clean_kp_train_loss": [
+    0.6793887242517973,
+    0.5070055065029546,
+    0.4780606180743167,
+    0.46464681211270786,
+    0.45651326932405173,
+    0.4414772361203244,
+    0.4295881901916705,
+    0.4249380529554267,
+    0.4178563620542225,
+    0.40975865175849513,
+    0.40546370631770084,
+    0.3982252176811821,
+    0.3858343799490678,
+    0.379007538933503,
+    0.3656208556576779,
+    0.35379995308424295,
+    0.33681938353337737,
+    0.32208908896697197,
+    0.30617407221543164,
+    0.2926546273419732,
+    0.28012341188757045,
+    0.27170656586948194,
+    0.2618562983211718,
+    0.25525567559819473,
+    0.24340780433855558,
+    0.241100468886526,
+    0.22936643666342685,
+    0.22153453873960596,
+    0.2213413413574821,
+    0.21846851916689622,
+    0.2104246342495868,
+    0.20888692896617087,
+    0.19663872825471979,
+    0.19428577112524134,
+    0.1947261790539089,
+    0.18938347912148426,
+    0.1853264942294673,
+    0.17872829263147555,
+    0.17811942146012658,
+    0.1693246652264344
+  ],
+  "gate": "fail",
+  "holdout_examples": 4096,
+  "model_seed": 4101,
+  "provenance": {
+    "cuda_visible_devices": "0",
+    "device": "cuda:0",
+    "git_commit": "b5eab6cd911e074ae9268303389dfbe52e636e26",
+    "torch_version": "2.3.1+cu118"
+  },
+  "raw_sdil_rollout_or_test_outcomes_read": false,
+  "ridge_relative_to_mean_soma_variance": 0.001,
+  "shuffle_seed": 4101,
+  "stage": "babyai_population_predictor_capacity_screen"
+}
-- 
cgit v1.2.3