Update README.md
Browse files
README.md
CHANGED
|
@@ -23,19 +23,55 @@ task_ids: []
|
|
| 23 |
|
| 24 |

|
| 25 |
|
| 26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
| 28 |
-
#
|
|
|
|
| 29 |
|
| 30 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
|
| 32 |
-
|
|
|
|
|
|
|
| 33 |
|
| 34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
-
##
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
-
|
|
|
|
| 39 |
|
| 40 |
### Domain
|
| 41 |
- Hr (48%)
|
|
|
|
| 23 |
|
| 24 |

|
| 25 |
|
| 26 |
+
---
|
| 27 |
+
license: cc-by-4.0
|
| 28 |
+
language:
|
| 29 |
+
- en
|
| 30 |
+
- hi
|
| 31 |
+
task_categories:
|
| 32 |
+
- text-classification
|
| 33 |
+
tags:
|
| 34 |
+
- hr
|
| 35 |
+
- compliance
|
| 36 |
+
- legal
|
| 37 |
+
- india
|
| 38 |
+
- united-states
|
| 39 |
+
- autoscientist-challenge-2026
|
| 40 |
+
pretty_name: HR Interview Question Compliance (Second Training Set)
|
| 41 |
+
size_categories:
|
| 42 |
+
- n<1K
|
| 43 |
+
---
|
| 44 |
|
| 45 |
+
# HR Interview Question Compliance — Second Training Dataset
|
| 46 |
+
### (`interview_question_samples`)
|
| 47 |
|
| 48 |
+
## Dataset Summary
|
| 49 |
+
A 40-row hand-written dataset of interview questions labeled for legal/bias risk across
|
| 50 |
+
India and the US, used as a second training set for
|
| 51 |
+
`adaption_llama_4_scout_17b_16_interview_question_sampl_1d90ba56`
|
| 52 |
+
(Llama-4-Scout-17B-16E-Instruct, LoRA) via Adaption's AutoScientist. Hand-written rather
|
| 53 |
+
than synthetically expanded, this set adds high-precision, human-verified examples
|
| 54 |
+
alongside the larger synthetically-expanded primary training set
|
| 55 |
+
(`hiring_question_compliance`).
|
| 56 |
|
| 57 |
+
All generalization and benchmark claims for this project are reported against a separate,
|
| 58 |
+
independently held-out evaluation set (`eval_set_v2.jsonl`, 30 rows, zero overlap with any
|
| 59 |
+
training data) — see that dataset's card for details.
|
| 60 |
|
| 61 |
+
## Schema
|
| 62 |
+
Same schema as the primary training dataset (`hiring_question_compliance`):
|
| 63 |
+
`job_description`, `question`, `jurisdiction`, `risk_label`, `risk_category`,
|
| 64 |
+
`legal_basis`, `justification`, `reframe`, plus an `id` field and a `flip_case_group`
|
| 65 |
+
field marking rows that form deliberate India/US jurisdiction flip-case pairs (identical
|
| 66 |
+
question, different correct verdict by jurisdiction).
|
| 67 |
|
| 68 |
+
## Composition
|
| 69 |
+
- 40 rows, 19 US / 21 India (approx.)
|
| 70 |
+
- Includes 5 jurisdiction flip-case pairs
|
| 71 |
+
- Same legal reference table grounding as the primary training dataset
|
| 72 |
|
| 73 |
+
## License
|
| 74 |
+
CC BY 4.0
|
| 75 |
|
| 76 |
### Domain
|
| 77 |
- Hr (48%)
|