Datasets:

HumanEdgeAI
/

LegalReasoning

@@ -1,91 +1,318 @@
 ---
-license: other
 language:
 - en
-tags:
-- legal
-- law
-- corporate-law
-- mergers-and-acquisitions
-- evaluation
-- rubrics
-- llm-evaluation
-pretty_name: Legal Tasks Sample Set (Tasks 89 & 93)
 size_categories:
 - n<1K
 task_categories:
 - text-generation
 - question-answering
 ---
-# Legal Tasks Sample Set (Tasks #89 & #93)
-## Dataset Summary
-This is a small, curated sample set of expert-authored **legal reasoning tasks** used to evaluate the quality of long-form legal writing (for example, responses produced by language models or by trainees). It contains **two tasks**, both in the **Corporate & M&A** area, each provided in three progressively refined versions (an original plus two revisions).
-Every task pairs a realistic legal prompt with a "gold" model answer written by a subject-matter expert, a detailed scoring rubric, reference materials, and metadata describing how realistic and how difficult the task is. The dataset is intended for evaluation and benchmarking rather than for training.
 ## Contents
-| Task | Area | Jurisdiction | Topic |
-|------|------|-------------|-------|
-| Legal Task #89 | Corporate & M&A | Belgium | Exclusion of a shareholder/CEO under Belgian company law |
-| Legal Task #93 | Corporate & M&A | England & Wales | Share buybacks under the Companies Act 2006 |
-## Data Structure
-The dataset is a single JSON object. Each top-level key is a task name (`"Legal Task #89"`, `"Legal Task #93"`). Each task contains an `Area` label and three versions:
-- **Original Version** — the first draft of the task
-- **Version 1** — a first revision
-- **Version 2** — a further revision
-Each version is an object with the following fields:
 | Field | Type | Description |
-|-------|------|-------------|
-| `Prompt` | string | The legal question or instruction posed to the writer/model |
-| `Gold Response` | string | The expert-authored reference answer |
-| `Associated Rubrics` | integer | The number of rubric criteria associated with the task |
-| `Rubrics` | list | The scoring rubric — a list of criteria (see below) |
-| `Task Details` | string | Notes on what the task involves and why |
-| `Reference Materials` | string | Source laws, codes, or links used for the answer |
-| `Realistic Task` | integer | Rating (1–5) of how realistic the task is |
-| `Realistic Explanation` | string | Rationale for the realism rating |
-| `Difficult Task` | integer | Rating (1–5) of how difficult the task is |
-| `Difficulty Explanation` | string | Rationale for the difficulty rating |
-### Rubric format
-Each entry in `Rubrics` is an object with these fields:
-- `rubric_category` — the dimension being scored, one of: *Structure & Style Rubric*, *Substance Rubric - Explicit requirements*, *Substance Rubric - Implicit expectations*, *Substance Rubric - Legal correctness*, *Substance Rubric - Reasoning quality*, *Sources and References Rubric*, and *Negative Rubric*.
-- `score_option` — the points available for meeting the criterion.
-- `criterion` — the specific thing being evaluated.
-- `justification` — why the criterion matters.
-## Intended Uses
-This sample set is designed for:
-- Evaluating and benchmarking the quality of AI-generated or human-written legal analysis
-- Studying rubric-based scoring of long-form legal responses
-- Illustrating how legal evaluation tasks are constructed and refined across versions
-It is **not** intended as a source of legal advice. The prompts and responses are illustrative legal analyses for evaluation purposes and should not be relied upon for any real legal matter.
-## Jurisdictions Covered
-- **Belgium** — Belgian Code of Companies and Associations (CCA) and Belgian Judicial Code (Task #89)
-- **England & Wales** — Companies Act 2006 (Task #93)
-## Limitations and Notes
-- This is a **two-task sample**, not a full dataset; it is meant for demonstration and small-scale evaluation.
-- Content is specific to the corporate-law scenarios described and does not generalize to other legal areas or jurisdictions.
-- Because the file is a single nested JSON object (keyed by task name) rather than a flat table of rows, the Hugging Face Dataset Viewer may not render it as a table. The file remains valid, downloadable, and machine-readable. A JSON Lines (`.jsonl`) version can be produced if a table view is required.
-## Citation
-If you use this sample set, please credit the original task authors and this repository.

 ---
+license: cc-by-nc-4.0
 language:
 - en
+pretty_name: 'Human Edge Legal Reasoning Evaluation — Showcase Sample'
 size_categories:
 - n<1K
 task_categories:
 - text-generation
 - question-answering
+tags:
+- legal
+- law
+- legal-reasoning
+- evaluation
+- rubric
+- rubrics
+- reward-model
+- rlhf
+- rlvr
+- expert-annotation
+- benchmark
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: legal_reasoning_showcase.parquet
+dataset_info:
+  features:
+  - name: task_id
+    dtype: int32
+  - name: area
+    dtype: string
+  - name: prompt
+    dtype: string
+  - name: golden_answer
+    dtype: string
+  - name: associated_rubrics
+    dtype: int32
+  - name: rubrics
+    list:
+    - name: rubric_category
+      dtype: string
+    - name: score_option
+      dtype: int32
+    - name: criterion
+      dtype: string
+    - name: justification
+      dtype: string
+  - name: task_details
+    dtype: string
+  - name: reference_materials
+    dtype: string
+  - name: realistic_explanation
+    dtype: string
+  - name: difficulty_explanation
+    dtype: string
+  - name: peer_review_1_overall_quality_label
+    dtype: string
+  - name: peer_review_1_overall_quality_description
+    dtype: string
+  - name: peer_review_1_representativeness_label
+    dtype: string
+  - name: peer_review_1_difficulty_label
+    dtype: string
+  - name: peer_review_1_difficulty_description
+    dtype: string
+  - name: peer_review_1_task_specification_quality_label
+    dtype: string
+  - name: peer_review_2_overall_quality_label
+    dtype: string
+  - name: peer_review_2_overall_quality_description
+    dtype: string
+  - name: peer_review_2_representativeness_label
+    dtype: string
+  - name: peer_review_2_difficulty_label
+    dtype: string
+  - name: peer_review_2_difficulty_description
+    dtype: string
+  - name: peer_review_2_task_specification_quality_label
+    dtype: string
+  splits:
+  - name: train
+    num_examples: 6
 ---
+# Human Edge — Legal Reasoning Evaluation (Showcase Sample)
+A public 6-task sample from a rubric-based legal reasoning evaluation dataset built by
+**Human Edge** ([humanedgetech.ai](https://humanedgetech.ai)). Each task is authored and reviewed by
+practicing senior lawyers and is designed to produce a **verifiable, per-criterion reward signal**
+for post-training and evaluation of frontier language models on high-complexity legal work.
+The sample contains **one task per legal subdomain**, drawn from a larger internal corpus of
+50 expert-authored tasks.
+## Why this dataset exists
+Most legal evaluations reduce to multiple-choice recall or bar-exam-style questions. Real legal work
+does not look like that: it is open-ended, multi-jurisdictional, and judged on reasoning quality,
+appropriate hedging, and whether the sources actually say what the answer claims they say.
+This dataset takes the opposite approach. Every task is a realistic instruction from a practicing
+lawyer's desk, paired with an expert-authored reference answer and a **fine-grained rubric of
+weighted, binary criteria** (20–40 across the source corpus).
+The rubric — not the reference answer — is the object of
+evaluation. That makes scoring reproducible across graders and directly convertible into
+per-dimension reward signals.
+## Task architecture
+Each row is a triple-component unit:
+| Component | Field | What it is |
+|---|---|---|
+| 1. Legal reasoning prompt | `prompt` | A realistic scenario requiring multi-step reasoning and professional judgment, with role, jurisdiction, and deliverable format specified |
+| 2. Golden answer | `golden_answer` | An expert-authored reference analysis (17k–38k characters in this sample) |
+| 3. Scoring rubric | `rubrics` | 20–40 categorized, weighted, binary criteria, each with the author's justification |
+## Rubric design
+Criteria are binary and weighted. Positive weights reward required content; negative weights
+penalize specific failure modes (hallucinated authority, overconfident advice, missing caveats).
+Criteria fall into four categories, with Substance further divided into four sub-categories:
+| Category | Sub-category | Criteria | What it tests |
+|---|---|---|---|
+| **Substance** | Explicit requirements | 27 | Did the answer do what was literally asked? |
+| | Implicit requirements | 27 | Did it surface what a competent practitioner would raise unprompted? |
+| | Legal correctness | 21 | Is the law stated accurately? |
+| | Reasoning quality | 31 | Is the analytical path sound, not just the conclusion? |
+| **Sources & References** | — | 34 | Are cited authorities real, relevant, and correctly characterized? |
+| **Structure & Style** | — | 30 | Does it read like professional work product? |
+| **Negative Criteria** | — | 25 | Penalties for specific, anticipated failure modes |
+| **Total** | | **195** | |
+The `rubric_category` field stores each of these as a single literal string. Filter on these values
+exactly:
+`Substance Rubric - Explicit requirements`, `Substance Rubric - Implicit requirements`,
+`Substance Rubric - Legal correctness`, `Substance Rubric - Reasoning quality`,
+`Sources and References Rubric`, `Structure & Style Rubric`, `Negative Rubric`
+Weight distribution across the 195 criteria in this sample:
+`+5` (69), `+3` (70), `+1` (31), `-1` (2), `-3` (5), `-5` (18).
 ## Contents
+Six tasks, one per subdomain:
+| `task_id` | `area` | Rubric criteria | Peer review overall quality (R1 / R2) |
+|---|---|---|---|
+| 18 | Employment & Labor | 34 | 3 / 4 |
+| 40 | Intellectual Property (IP) | 40 | 4 / 4 |
+| 68 | Commercial Litigation | 34 | 4 / 5 |
+| 82 | Regulatory & Compliance | 29 | 4 / 5 |
+| 93 | Corporate & M&A | 30 | 4 / 4 |
+| 100 | Contract Law | 28 | 4 / 5 |
+Jurisdictions are US and UK (England and Wales). `task_id` values are the original corpus
+identifiers and are intentionally non-contiguous.
+## Who wrote and reviewed these tasks
+Tasks were authored by senior legal practitioners recruited against a hard credential bar:
+8+ years in practice, a Master's or PhD in law, and a background in AmLaw 100 / Magic Circle firms,
+senior courts, government, or in-house at large enterprises.
+The contributing cohort averaged **~18.5 years of practice**, split roughly 75% US / 25% UK,
+with education spanning T14 US law schools and leading UK and European universities. Identity was
+verified for every contributor before any project interaction.
+Domain expertise alone does not make a calibrated evaluator, so every contributor completed a
+mandatory training program (~2.3 hours average) on stress-testing model outputs, applying rubrics
+consistently, and documenting rationale.
+## Quality assurance
+Three stages, applied to every task in the source corpus:
+1. **Automated checks.** Every task passes programmatic validation before it reaches a reviewer.
+   Prompts are checked for scenario framing, role specification, deliverable format, and
+   jurisdictional context. Rubrics are checked for binary format compliance and a minimum criterion
+   count. A model-graded pass confirms each rubric is consistent with its golden answer.
+2. **Double-blind expert peer review.** Two independent senior practitioners review each task,
+   rating overall quality, difficulty, representativeness, and task specification quality, each with
+   written rationale. Reviews drive iterative revision rather than a simple accept/reject: the author
+   revises against reviewer comments, and the version shipped here reflects those revisions.
+3. **Difficulty validation.** Tasks are stress-tested against frontier models. A task is admitted
+   only if strong models still fail a meaningful share of its rubric criteria — tasks that models
+   solve comfortably carry no training signal and are rejected.
+Tasks in the source corpus averaged roughly eleven hours of expert effort each, counting authoring,
+peer review, and revision. Reviewer ratings clustered tightly, indicating consensus among
+practitioners rather than averaged-out disagreement.
+## Field reference
+### Task content
+| Field | Type | Description |
+|---|---|---|
+| `task_id` | int32 | Original corpus identifier |
+| `area` | string | Legal subdomain |
+| `prompt` | string | The legal reasoning prompt given to the model |
+| `golden_answer` | string | Expert-authored reference answer |
+| `associated_rubrics` | int32 | Number of rubric criteria; always equals `len(rubrics)` |
+| `rubrics` | list&lt;struct&gt; | Nested rubric criteria (see below) |
+| `task_details` | string | Author's note on why the task is hard or interesting |
+| `reference_materials` | string | Source authorities relied on; free-text, sometimes URLs, sometimes citations |
+### `rubrics` struct
 | Field | Type | Description |
+|---|---|---|
+| `rubric_category` | string | One of the seven literal values listed above |
+| `score_option` | int32 | Weight: `+5`, `+3`, `+1`, `-1`, `-3`, `-5` |
+| `criterion` | string | Binary question applied to the model's answer |
+| `justification` | string | Author's rationale for why this criterion matters |
+Text has been normalized: non-breaking spaces, stray tabs and other exotic whitespace introduced
+by the authoring tools have been folded to plain spaces. Newlines in `golden_answer` are preserved.
+### Author commentary
+The task author's own notes on realism and difficulty. Their numeric self-ratings are deliberately
+**not** included — an author's rating of their own task is not independent evidence, so the only
+ratings in this dataset come from the two peer reviewers.
+| Field | Type | Description |
+|---|---|---|
+| `realistic_explanation` | string | Why the scenario reflects real practice |
+| `difficulty_explanation` | string | What makes the task hard, and where models are expected to fail |
+### Peer review
+Two independent reviewers per task, both prefixed: `peer_review_1_` is the first-round review
+(conducted on the first draft version), `peer_review_2_` the second (conducted on the second draft
+version). The final version shipped here was produced after revising comments from the second
+reviewer. Reviewer identities are not published.
+The `*_label` fields carry the rating exactly as the reviewer selected it — score and wording
+together, e.g. `4 - Above Standards`. To get a numeric value, split on the first `-`.
+| Field suffix | Type | Description |
+|---|---|---|
+| `overall_quality_label` | string | Rating 1–5, e.g. `4 - Above Standards` |
+| `overall_quality_description` | string | Reviewer's written rationale for the quality rating |
+| `representativeness_label` | string | Rating 1–5 — how typical this task is of real practice, e.g. `5 - Core Activity` |
+| `difficulty_label` | string | Rating 1–5, e.g. `3 - Moderate` |
+| `difficulty_description` | string | Reviewer's written rationale for the difficulty rating |
+| `task_specification_quality_label` | string | Categorical: `Well-Specified`, `Overspecified`, or `Underspecified` |
+Rating scales, as presented to reviewers:
+- **Overall quality:** `1 - Far Below Standards`, `2 - Below Standards`, `3 - Meets Minimum Standards`, `4 - Above Standards`, `5 - Exemplary`
+- **Representativeness:** `1 - Rarely Encountered`, `2 - Uncommon`, `3 - Somewhat Typical`, `4 - Typical`, `5 - Core Activity`
+- **Difficulty:** `1 - Trivial` → `5 - Very Hard` (`3 - Moderate`, `4 - Hard`)
+- **Task specification quality:** `Underspecified`, `Well-Specified`, `Overspecified`
+## Intended uses
+- Evaluating LLM performance on open-ended, high-complexity legal reasoning
+- Rubric-as-reward research: RLVR, reward model training, process supervision
+- LLM-as-judge calibration — the rubrics give a judge concrete, verifiable criteria instead of a
+  vague quality prompt
+- Studying legal-domain failure modes, especially source and citation reliability
+- A reference template for anyone constructing expert-authored rubric evaluations in other
+  professional domains
+## Limitations
+- **Six tasks.** This is a methodology sample, not a benchmark. Do not report aggregate scores over
+  six tasks as a model capability claim.
+- **Scoring requires a judge.** Criteria are binary but not mechanically checkable; they need a
+  competent grader, human or model. Judge choice will shift absolute scores.
+- **US and UK only.** Nothing here generalizes to civil law systems, EU-level practice, or other
+  common law jurisdictions without revalidation.
+- **Contamination risk.** Published openly, these tasks may enter future training corpora. Treat
+  results on this sample as indicative once models trained after its publication are involved.
+- **Reviewer identities are withheld.** Cohort-level credentials are described above; individual
+  names and affiliations are not published.
+- **Not legal advice.** The golden answers are evaluation artifacts written against hypothetical
+  facts. They are not advice and must not be relied on.
+## About Human Edge
+Human Edge builds audit-grade expert judgment infrastructure for AI labs: sourcing credentialed
+professionals, training them as calibrated evaluators, and producing rubric-based evaluation data
+with full provenance — who judged, their qualifications, their rationale, and the revision history
+behind every task.
+Contact: [humanedgetech.ai](https://humanedgetech.ai)
+## License
+Released under [**CC BY-NC 4.0**](https://creativecommons.org/licenses/by-nc/4.0/)
+(Creative Commons Attribution-NonCommercial 4.0 International).
+**You may**, at no cost:
+- Use the dataset to evaluate, benchmark, and audit models
+- Share and redistribute it, in whole or in part
+- Adapt it — build derived rubrics, translate it, extend the schema — and publish those adaptations
+**Conditions:**
+- **Attribution.** Credit Human Edge and link back to this dataset.
+- **NonCommercial.** Commercial use requires a separate license. This includes training or
+  fine-tuning models that are sold or offered as a paid service.
+**Commercial use, including model training:** contact Human Edge at
+[humanedgetech.ai](https://humanedgetech.ai). Commercial terms are readily available, and we
+license the full corpus and custom expert-authored datasets in other professional domains.
+Nothing in this license grants rights in the third-party statutes, regulations, and court opinions
+cited within the tasks; those remain governed by their own terms.