Update README.md
Browse files
README.md
CHANGED
|
@@ -5,8 +5,7 @@ pretty_name: TraceBench
|
|
| 5 |
|
| 6 |
# TraceBench
|
| 7 |
|
| 8 |
-
This dataset contains the
|
| 9 |
-
|
| 10 |
The matching evaluation and reproduction code is available at [TommasoBendinelli/TraceBench](https://github.com/TommasoBendinelli/TraceBench).
|
| 11 |
|
| 12 |
## Layout
|
|
@@ -15,20 +14,4 @@ The matching evaluation and reproduction code is available at [TommasoBendinelli
|
|
| 15 |
- `results.csv` and `results.parquet` contain the same 420 aggregate result rows and 47 columns.
|
| 16 |
- `submissions/` contains the four accepted submission bundles rebuilt from the exact 420 source trajectories selected by the paper.
|
| 17 |
- `website/` contains the static website data.
|
| 18 |
-
- `release_manifest.json` records hashes and row coverage for the synchronized release.
|
| 19 |
-
|
| 20 |
-
## Results provenance
|
| 21 |
-
|
| 22 |
-
Accuracy and resource metrics are synchronized to the paper's 420-row execution record, which contains 360 `PAPER` rows and 60 `CALIBRATION` rows.
|
| 23 |
-
|
| 24 |
-
The four deprecated `llm_*` classifier columns were removed, and the paper-backed `python_statements`, `total_time_seconds`, `elapsed_time_seconds`, `session_limit_wait_seconds`, and `total_time_method` fields were added.
|
| 25 |
-
|
| 26 |
-
Console-output shares use the paper's raw-ATIF recalculation for the 240 main-condition rows, while the remaining trajectory metrics use the authoritative evaluation selected for each paper row.
|
| 27 |
-
|
| 28 |
-
GPT-5.5 costs use the standard text-token rates of $5.00 input, $0.50 cached input, and $30.00 output per million tokens from the [official OpenAI documentation](https://developers.openai.com/api/docs/models/gpt-5.5).
|
| 29 |
-
|
| 30 |
-
## Privacy and packaging
|
| 31 |
-
|
| 32 |
-
Machine-local result path columns are excluded from both aggregate result files.
|
| 33 |
|
| 34 |
-
Generated bytecode, intermediate raw exports, and stale duplicate archives are excluded from this release.
|
|
|
|
| 5 |
|
| 6 |
# TraceBench
|
| 7 |
|
| 8 |
+
This dataset contains the public benchmark data, synchronized aggregate results, trajectories, website data, and accepted submission artifacts for TraceBench.
|
|
|
|
| 9 |
The matching evaluation and reproduction code is available at [TommasoBendinelli/TraceBench](https://github.com/TommasoBendinelli/TraceBench).
|
| 10 |
|
| 11 |
## Layout
|
|
|
|
| 14 |
- `results.csv` and `results.parquet` contain the same 420 aggregate result rows and 47 columns.
|
| 15 |
- `submissions/` contains the four accepted submission bundles rebuilt from the exact 420 source trajectories selected by the paper.
|
| 16 |
- `website/` contains the static website data.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
|
|