Update readme
Browse files
versions/iclr2027-rebuttal-v1/README.md
CHANGED
|
@@ -8,10 +8,10 @@ This collection contains two separate, anonymous artifacts for the revised bench
|
|
| 8 |
|
| 9 |
| Archive | Contents | Size (bytes) |
|
| 10 |
|---|---|---:|
|
| 11 |
-
| `SparseWake_parameter_evaluation_extension_v1.zip` | Additional raw observation histories, targets, sensor coordinates, condition metadata, seed-specific split eligibility, saved predictions and scoring code for observer probe size, relative motion, and source speed under two controlled protocols. | 76,589,
|
| 12 |
-
| `SparseWake_revision_evidence_v1.zip` | Geometry-distance diagnostics, physics-based inverse search evidence, joint source-set models and predictions, frozen-encoder point/Gaussian/mixture heads and predictions, failure-category summaries, protocols and replay/reproduction code. | 192,
|
| 13 |
|
| 14 |
-
The archives together occupy **268,
|
| 15 |
|
| 16 |
### Controlled parameter evaluation
|
| 17 |
|
|
@@ -23,7 +23,7 @@ This is an **evaluation-only** extension of the canonical single-source task. It
|
|
| 23 |
|
| 24 |
The evidence archive includes six saved multi-hypothesis prediction files and nine trained output heads (point, Gaussian, and mixture) on the three frozen MLP representations. It also includes twelve joint source-set checkpoints, saved source-set predictions, inverse search records, new confirmation observations, and the retained diagnostic sufficient statistics. Machine-specific path metadata in release copies of checkpoints and protocols was replaced with neutral identifiers. Numeric arrays and tensor fields were preserved and checked against the originals.
|
| 25 |
|
| 26 |
-
The source-set models use known source count and all-source position supervision. The output heads retain nearest-source supervision on frozen representations. These distinctions remain part of the experiment definitions. The archive provides evidence and metric replay, plus deeper reproduction entry points with stated dependencies.
|
| 27 |
|
| 28 |
## Existing public inputs
|
| 29 |
|
|
@@ -49,8 +49,8 @@ python revision_evidence/replay_metrics.py --out revision_metric_replay.json
|
|
| 49 |
python revision_evidence/replay_metrics.py --core PATH_TO_EXTRACTED_CORE --out revision_metric_replay_all.json
|
| 50 |
```
|
| 51 |
|
| 52 |
-
Write replay outputs outside the extracted artifact directories. `revision_evidence/geometry/independent_check.py`, `multi_hypothesis/REPRODUCE.md` and `coauthor_code/README.md` describe optional deeper checks, environments and entry points. The default replay
|
| 53 |
|
| 54 |
Independent preparation checks verified ZIP integrity and every archive member hash. Parameter metrics replayed exactly. The combined revision replay checked 674 numerical fields, with maximum absolute difference approximately 1.42e-14. Those checks validate numerical consistency of the retained evidence.
|
| 55 |
|
| 56 |
-
|
|
|
|
| 8 |
|
| 9 |
| Archive | Contents | Size (bytes) |
|
| 10 |
|---|---|---:|
|
| 11 |
+
| `SparseWake_parameter_evaluation_extension_v1.zip` | Additional raw observation histories, targets, sensor coordinates, condition metadata, seed-specific split eligibility, saved predictions and scoring code for observer probe size, relative motion, and source speed under two controlled protocols. | 76,589,015 |
|
| 12 |
+
| `SparseWake_revision_evidence_v1.zip` | Geometry-distance diagnostics, physics-based inverse search evidence, joint source-set models and predictions, frozen-encoder point/Gaussian/mixture heads and predictions, failure-category summaries, protocols and replay/reproduction code. | 192,032,988 |
|
| 13 |
|
| 14 |
+
The archives together occupy **268,622,003 bytes** (268.62 MB, 256.18 MiB). SHA-256 checksums are in `SHA256SUMS.txt`; `UPLOAD_MANIFEST.json` describes the public files using relative names only. Each archive also contains its own member-level manifest.
|
| 15 |
|
| 16 |
### Controlled parameter evaluation
|
| 17 |
|
|
|
|
| 23 |
|
| 24 |
The evidence archive includes six saved multi-hypothesis prediction files and nine trained output heads (point, Gaussian, and mixture) on the three frozen MLP representations. It also includes twelve joint source-set checkpoints, saved source-set predictions, inverse search records, new confirmation observations, and the retained diagnostic sufficient statistics. Machine-specific path metadata in release copies of checkpoints and protocols was replaced with neutral identifiers. Numeric arrays and tensor fields were preserved and checked against the originals.
|
| 25 |
|
| 26 |
+
The source-set models use known source count and all-source position supervision. The output heads retain nearest-source supervision on frozen representations. These distinctions remain part of the experiment definitions. The archive provides evidence and metric replay, plus deeper reproduction entry points with stated dependencies.
|
| 27 |
|
| 28 |
## Existing public inputs
|
| 29 |
|
|
|
|
| 49 |
python revision_evidence/replay_metrics.py --core PATH_TO_EXTRACTED_CORE --out revision_metric_replay_all.json
|
| 50 |
```
|
| 51 |
|
| 52 |
+
Write replay outputs outside the extracted artifact directories. `revision_evidence/geometry/independent_check.py`, `multi_hypothesis/REPRODUCE.md` and `coauthor_code/README.md` describe optional deeper checks, environments and entry points. The default replay recalculates the reported metrics from stored experiment outputs. Full experiment entry points are described in the linked reproduction guides.
|
| 53 |
|
| 54 |
Independent preparation checks verified ZIP integrity and every archive member hash. Parameter metrics replayed exactly. The combined revision replay checked 674 numerical fields, with maximum absolute difference approximately 1.42e-14. Those checks validate numerical consistency of the retained evidence.
|
| 55 |
|
| 56 |
+
The original core release supplies the base observations and encoders; the two extension archives supply parameter-evaluation observations and the additional model experiments described above.
|