living pointer: GET https://councilof.ai/api/gspc
Browse files
README.md
CHANGED
|
@@ -1,65 +1,9 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
-
tags:
|
| 4 |
-
- csoai
|
| 5 |
-
- gspc
|
| 6 |
-
- governance
|
| 7 |
-
- safety
|
| 8 |
-
- evaluation
|
| 9 |
-
- benchmark
|
| 10 |
---
|
| 11 |
|
| 12 |
-
#
|
| 13 |
|
| 14 |
-
|
| 15 |
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
Every row has exactly these fields (one unified schema for all sources):
|
| 19 |
-
|
| 20 |
-
| field | type | description |
|
| 21 |
-
|---|---|---|
|
| 22 |
-
| `source` | string | Original HF dataset id (`csoai/gspc-*`) |
|
| 23 |
-
| `axis` | string | Short axis id (`gov`, `prv`, `asi`, ...) |
|
| 24 |
-
| `category` | string | Original dataset's category/axis label (may be empty if source had none) |
|
| 25 |
-
| `anchor` | string | Original dataset's anchor/note/source/why field (may be empty) |
|
| 26 |
-
| `prompt` | string | The question/scenario text (the input to the model) |
|
| 27 |
-
| `expected` | string | The expected answer — string-scalar for classification, ` || `-joined for multi-label |
|
| 28 |
-
| `expected_is_list` | bool | True if original gold was a list (multi-label source) |
|
| 29 |
-
| `raw_keys` | string | JSON dump of the original row's field names (debug aid) |
|
| 30 |
-
|
| 31 |
-
## Sources mapped
|
| 32 |
-
|
| 33 |
-
The underlying `csoai/gspc-*` datasets used six different field-name conventions for the prompt and gold fields. This dataset unifies them.
|
| 34 |
-
|
| 35 |
-
| Source | axis | rows |
|
| 36 |
-
|---|---|---|
|
| 37 |
-
| `csoai/gspc-agi` | agi | 36 |
|
| 38 |
-
| `csoai/gspc-art5` | art5 | 36 |
|
| 39 |
-
| `csoai/gspc-asi` | asi | 33 |
|
| 40 |
-
| `csoai/gspc-det` | det | 6 |
|
| 41 |
-
| `csoai/gspc-gov` | gov | 237 |
|
| 42 |
-
| `csoai/gspc-mach` | mach | 33 |
|
| 43 |
-
| `csoai/gspc-mcp` | mcp | 35 |
|
| 44 |
-
| `csoai/gspc-oss` | oss | 32 |
|
| 45 |
-
| `csoai/gspc-prv` | prv | 32 |
|
| 46 |
-
| `csoai/gspc-swarm` | swarm | 6 |
|
| 47 |
-
| `csoai/gspc-xr` | xr | 32 |
|
| 48 |
-
|
| 49 |
-
## Skipped
|
| 50 |
-
|
| 51 |
-
- **`csoai/gspc-care`** — intentionally excluded. Its rows have only metadata (`battery`, `timestamp`, `v2`) with no prompt field. Including it would require a manual annotation pass to add prompts. The source dataset itself remains untouched on HF.
|
| 52 |
-
|
| 53 |
-
## Provenance
|
| 54 |
-
|
| 55 |
-
Built 2026-08-08 by `gspc_normalize.py` after an external benchmark spray revealed the schema drift across the corpus. The unifying schema makes the gspc-* corpus usable by downstream benchmark runners, fine-tuners, and eval harnesses without per-dataset field-mapping code.
|
| 56 |
-
|
| 57 |
-
## License
|
| 58 |
-
|
| 59 |
-
MIT. Original datasets were MIT per the csoai org convention.
|
| 60 |
-
|
| 61 |
-
## Intended use
|
| 62 |
-
Measurement item bank / results for the GSPC 14-slot board (live: 14 measured of 14). Consumed by the estate's deterministic harness and graded with published predicates (unparsed counts incorrect). Not a training corpus.
|
| 63 |
-
|
| 64 |
-
## Limitations
|
| 65 |
-
Bank scope and n are stated above; unmeasured cells are stated, never interpolated. Items target governance-rule compliance, not general capability. Scores are point-in-time — the live board at https://councilof.ai/api/gspc supersedes any static copy.
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
| 4 |
|
| 5 |
+
# gspc-normalized
|
| 6 |
|
| 7 |
+
Council of AI measurement bank. Measurement, not certification.
|
| 8 |
|
| 9 |
+
GET https://councilof.ai/api/gspc
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|