docs: synchronize corrected Unified E5 results

INTACT-unified/AUDITED_RESULTS_20260914.json

@@ -0,0 +1,52 @@
+{
+  "schema_version": 1,
+  "result_revision": "2026-09-14",
+  "criterion": "Official",
+  "protocol": {
+    "name": "corrected_causal_continuation",
+    "training_seeds": [0, 42, 3072],
+    "evaluation_seeds": [0, 1, 42],
+    "episodes_per_evaluation_seed": 100,
+    "history_boundary": "strictly pre-start actions only; append each policy-executed action exactly once"
+  },
+  "statistics": "mean and sample standard deviation across training-seed means",
+  "results_percent": [
+    {
+      "cell": "LeWM",
+      "inference": "CEM 300x30",
+      "pusht": {"mean": 74.56, "sample_std": 3.67},
+      "cube": {"mean": 67.33, "sample_std": 1.86},
+      "reacher": {"mean": 83.11, "sample_std": 0.96},
+      "tworoom": {"mean": 39.67, "sample_std": 8.39},
+      "macro": {"mean": 66.17, "sample_std": 2.67}
+    },
+    {
+      "cell": "Inverse only",
+      "inference": "Direct",
+      "pusht": {"mean": 36.11, "sample_std": 0.19},
+      "cube": {"mean": 67.56, "sample_std": 4.03},
+      "reacher": {"mean": 90.56, "sample_std": 1.35},
+      "tworoom": {"mean": 78.56, "sample_std": 6.68},
+      "macro": {"mean": 68.19, "sample_std": 0.67}
+    },
+    {
+      "cell": "Goal intent only",
+      "inference": "Direct",
+      "pusht": {"mean": 81.78, "sample_std": 0.96},
+      "cube": {"mean": 100.00, "sample_std": 0.00},
+      "reacher": {"mean": 88.67, "sample_std": 0.33},
+      "tworoom": {"mean": 69.22, "sample_std": 7.04},
+      "macro": {"mean": 84.92, "sample_std": 1.86}
+    },
+    {
+      "cell": "Goal-displacement INTACT",
+      "inference": "Direct",
+      "pusht": {"mean": 86.11, "sample_std": 0.96},
+      "cube": {"mean": 100.00, "sample_std": 0.00},
+      "reacher": {"mean": 97.22, "sample_std": 0.84},
+      "tworoom": {"mean": 81.56, "sample_std": 3.67},
+      "macro": {"mean": 91.22, "sample_std": 0.51}
+    }
+  ],
+  "manifest_note": "PAPER_E5_* manifests preserve checkpoint identities and hashes; this file is authoritative for corrected-evaluator success rates."
+}

INTACT-unified/README.md

@@ -9,29 +9,34 @@ PushT, Cube, Reacher, and TwoRoom, with task-specific lightweight heads.
 | Cell | Archive prefix | Native interface | Official macro SR |
 |---|---|---|---:|
 | LeWM | `paper-e5-lewm-*` | CEM 300x30 | 66.17 |
-| Inverse only | `paper-e5-inverse-only-*` | Direct | 56.67 |
-| Waypoint intent only | `paper-e5-waypoint-intent-only-*` | Direct | 74.22 |
-| Goal intent only | `paper-e5-goal-intent-only-*` | Direct | 80.61 |
-| Waypoint INTACT | `paper-e5-waypoint-intact-*` | Direct | 76.42 |
-| **Goal-displacement INTACT** | `intact-goal-e5-*` | **Direct** | **89.39** |
+| Inverse only | `paper-e5-inverse-only-*` | Direct | 68.19 |
+| Goal intent only | `paper-e5-goal-intent-only-*` | Direct | 84.92 |
+| **Goal-displacement INTACT** | `intact-goal-e5-*` | **Direct** | **91.22** |
 
 The Goal-displacement INTACT task-level result is:
 
 | PushT | Cube | Reacher | TwoRoom | Macro SR |
 |---:|---:|---:|---:|---:|
-| 80.22 +/- 1.26 | 99.56 +/- 0.77 | 95.67 +/- 1.76 | 82.11 +/- 4.11 | **89.39 +/- 0.77** |
+| 86.11 +/- 0.96 | 100.00 +/- 0.00 | 97.22 +/- 0.84 | 81.56 +/- 3.67 | **91.22 +/- 0.51** |
 
 Success rates are percentages averaged over training seeds `0/42/3072`.
 Each checkpoint is evaluated on 100 episodes for evaluation seeds `0/1/42`;
 `+/-` is the sample standard deviation across training-seed means.
 
+These values use the corrected causal-continuation evaluator: the initial
+actor history contains only strictly pre-start actions, and each action
+executed by the evaluated policy is appended exactly once. The current values
+are recorded in `AUDITED_RESULTS_20260914.json`.
+
 ## Contents
 
-- Six controlled cells and three training seeds.
+- Six archived training cells and three training seeds. The current comparison
+  above reports the four goal-displacement cells used by the revised analysis.
 - Four task-specific checkpoint shards in each archive.
 - `6 x 3 x 4 = 72` checkpoints in total.
-- `PAPER_E5_MATRIX_MANIFEST.json`: per-task results and extracted shard hashes.
-- `PAPER_E5_GOAL_MANIFEST.json`: Goal-displacement INTACT metadata and results.
+- `AUDITED_RESULTS_20260914.json`: authoritative corrected-evaluator SR values.
+- `PAPER_E5_MATRIX_MANIFEST.json`: checkpoint inventory and extracted shard hashes.
+- `PAPER_E5_GOAL_MANIFEST.json`: Goal-displacement INTACT checkpoint metadata.
 - `SHA256SUMS`: archive integrity records.
 - `RUNTIME_SHA256SUMS`: frozen evaluator runtime fingerprint.
 

README.md

@@ -62,8 +62,8 @@ shared-encoder comparison matrix.
 | Goal intent only | Direct | 84.92 |
 | **Goal-displacement INTACT** | **Direct** | **91.22** |
 
-All cells use the same E5 shared-encoder protocol. Task-level results,
-per-seed statistics, checkpoint hashes, and runtime fingerprints are provided
+All cells use the same E5 shared-encoder protocol. Task-level statistics,
+checkpoint hashes, and runtime fingerprints are provided
 inside [`INTACT-unified`](https://huggingface.co/INTACT-JEPA/INTACT/tree/main/INTACT-unified).
 The immutable original layout also remains available at the frozen
 [`paper-e5-goal-v1`](https://huggingface.co/INTACT-JEPA/INTACT/tree/paper-e5-goal-v1)