Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,204 @@
+---
+license: cc-by-4.0
+language:
+  - en
+pretty_name: "The Human Creativity Benchmark (HCB)"
+task_categories:
+  - text-to-image
+  - image-to-video
+  - text-generation
+  - other
+tags:
+  - creative-evaluation
+  - human-evaluation
+  - preference-data
+  - pairwise-comparison
+  - evaluator-agreement
+  - convergence-divergence
+  - creativity
+  - design
+  - text-to-image
+  - image-to-video
+  - code-generation
+  - benchmark
+size_categories:
+  - 1K<n<10K
+configs:
+  - config_name: prompts_workflow
+    data_files: prompts_workflow.csv
+  - config_name: model_outputs
+    data_files: model_outputs.csv
+  - config_name: pairwise_comparisons
+    data_files: pairwise_comparisons.csv
+  - config_name: scalar_feedback
+    data_files: scalar_feedback.csv
+  - config_name: qualitative_feedback
+    data_files: qualitative_feedback.csv
+---
+
+# The Human Creativity Benchmark (HCB)
+
+Expert evaluations of AI-generated creative work, built to separate two signals that single-score benchmarks collapse: **convergence**, where professionals align around shared, checkable standards, and **divergence**, where creative taste legitimately differs. Each AI output is judged by domain professionals through three complementary lenses — forced-choice pairwise comparisons, 1-5 scalar ratings on prompt adherence, usability, and visual appeal, and open-ended written rationale that is coded into themes and sentiment. The benchmark spans five creative domains (ad images, brand design, ad video, desktop apps, landing pages) and three workflow phases (Ideation, Mockup, Refinement), so a model can be assessed not just on whether an output is good, but good according to whom, for what purpose, and at what stage of the creative process.
+
+> **A sample of the human data [Contra Labs](https://contralabs.com/?utm_source=huggingface&utm_medium=dataset_card&utm_campaign=human-creativity-benchmark&utm_content=hero) builds on demand.** We are an independent human-data and creative-evaluation lab: expert evaluation, rankings, and benchmarks for AI outputs, plus custom datasets like this one. See [Working with Contra Labs](#working-with-contra-labs).
+
+This dataset accompanies the paper *The Human Creativity Benchmark: Studying Convergence and Divergence in an Expert-Labeled Benchmark* (Hopkins, Nulty, Minetti; preprint, June 2026).
+
+## Motivation
+
+Most AI evaluation treats evaluator disagreement as noise to be averaged away. In creative work that erases the most useful signal: professionals genuinely agree on verifiable dimensions (readable typography, working layout, correct visual hierarchy) and genuinely diverge on taste-driven ones (aesthetic direction, mood, conceptual risk). HCB preserves both. The intended use is to study where a model should be **reliably correct** versus where it should remain **steerable** to creative preference — and how that balance shifts across the arc of a project. A recurring finding in the accompanying paper: no model led all three workflow phases in any domain, so the data is as much about phase-dependent strengths as about overall quality.
+
+## What the dataset contains
+
+Domain professionals were drawn from a network of independent creatives and matched to the model category most relevant to their workflow, then asked to evaluate anonymized, randomly ordered model outputs at each phase. Prompts chain across phases to mimic a real designer's process: Mockup and Refinement prompts build on the prior phase and supply seed images where applicable.
+
+| | |
+|---|---|
+| Creative domains | 5 — Ad Images, Ad Video, Brand Design, Desktop App, Landing Pages |
+| Workflow phases | 3 — Ideation, Mockup, Refinement |
+| Prompts | 95 |
+| AI model outputs | 380 |
+| Models evaluated | 13 |
+| Evaluators | 31 (anonymized) |
+| Pairwise judgments | 3,174 |
+| Scalar evaluation rows | 2,247 (812 with numeric 1-5 ratings — see note) |
+| Qualitative feedback rows | 2,247 |
+
+### Domains, modalities, and models
+
+| Domain | Modality | Models |
+|---|---|---|
+| Ad Images | text-to-image / image-to-image | `gpt-image-1.5`, `gemini-3-pro-image-preview`, `seedream-4.5`, `flux-2-pro` |
+| Brand Design | text-to-image / image-to-image | `gpt-image-1.5`, `gemini-3-pro-image-preview`, `seedream-4.5`, `flux-2-max` |
+| Ad Video | image-to-video | `veo3.1`, `kling-v3.0-pro`, `seedance-v1.5-pro`, `grok-imagine-video` |
+| Desktop App | text-to-code / code-to-code | `claude-opus-4.6`, `gemini-3.1-pro-preview`, `gpt-5.3-codex`, `qwen3.5-397b-a17b` |
+| Landing Pages | text-to-code / code-to-code | `claude-opus-4.6`, `gemini-3.1-pro-preview`, `gpt-5.3-codex`, `qwen3.5-397b-a17b` |
+
+### Workflow phases
+
+- **Ideation** — discovery and exploration; the goal is exciting, strategically appropriate creative direction rather than final quality.
+- **Mockup** — the chosen direction is realized: product shots, scene composition, brand identity.
+- **Refinement** — near production-ready; targeted edits for consistency and polish.
+
+## Format and schema
+
+Five CSV files, joinable on shared keys. All tables join on `prompt_id`; `model_outputs.csv` adds `output_id` (`{prompt_id}__{model_id}`); the scalar and qualitative tables share `evaluation_id` (`{prompt_id}__{model_id}__{user_id}`). `domain` and `stage` can be recovered for any record from `prompts_workflow.csv` or `model_outputs.csv`.
+
+```python
+from datasets import load_dataset
+
+pairwise = load_dataset("contra-labs/HCB", "pairwise_comparisons", split="train")
+prompts  = load_dataset("contra-labs/HCB", "prompts_workflow", split="train")
+
+row = pairwise[0]
+print(row["domain"], row["stage"], "->", row["chosen_model"])
+```
+
+### `prompts_workflow.csv` — 95 rows (one per prompt)
+
+| Column | Description |
+|---|---|
+| `prompt_id` | Unique prompt identifier (e.g. `ad_image_idea_4152`). |
+| `domain` | One of the five creative domains. |
+| `stage` | `Ideation`, `Mockup`, or `Refinement`. |
+| `prompt_text` | The text prompt presented to the models. |
+| `input_image` | URL of the seed/reference image used as model input. Present mainly for Mockup and Refinement (44 of 95 rows); empty for most Ideation prompts. |
+
+### `model_outputs.csv` — 380 rows (one per prompt × model)
+
+| Column | Description |
+|---|---|
+| `output_id` | Unique output identifier (`{prompt_id}__{model_id}`). |
+| `prompt_id` | Foreign key to `prompts_workflow.csv`. |
+| `domain`, `stage` | Creative domain and workflow phase. |
+| `model_id` | Model that produced the output. |
+| `model_output` | The generated artifact. **Format depends on modality:** image domains (Ad Images, Brand Design) store a `cloudinary:<asset_id>` reference; code domains (Desktop App, Landing Pages) store the raw generated HTML/code inline; the Ad Video field is empty in this release (videos are not bundled in the CSV). |
+| `contestant_id` | Internal identifier for the output as presented in the evaluation interface. |
+
+### `pairwise_comparisons.csv` — 3,174 rows
+
+Forced-choice head-to-head preferences. For each prompt, evaluators compared all pairings of the four candidate models and selected the one they preferred; in the paper these aggregate via a Bradley-Terry model into ELO ratings by domain and phase.
+
+| Column | Description |
+|---|---|
+| `comparison_id` | Unique comparison identifier. |
+| `prompt_id` | Foreign key to `prompts_workflow.csv`. |
+| `domain`, `stage` | Creative domain and workflow phase. |
+| `user_id` | Anonymized evaluator identifier. |
+| `user_core_skill` | Evaluator's professional specialty (e.g. Brand Designer, Product Designer, Digital Marketer). |
+| `model_left`, `model_right` | The two models compared. |
+| `chosen_model` | The preferred model (always equal to `model_left` or `model_right`). |
+
+### `scalar_feedback.csv` — 2,247 rows
+
+1-5 Likert ratings on three dimensions, ordered from most objective to most taste-driven.
+
+| Column | Description |
+|---|---|
+| `evaluation_id` | Unique identifier (`{prompt_id}__{model_id}__{user_id}`). |
+| `prompt_id`, `user_id`, `user_core_skill`, `model_id` | Join keys and evaluator/model identifiers. |
+| `prompt_adherence` | 1-5: how faithful the output is to the prompt (least subjective). |
+| `usability` | 1-5: how well the output would function in a professional context. |
+| `visual_appeal` | 1-5: how visually interesting, cohesive, and polished it is (most taste-driven). |
+
+> **Note on coverage:** In this release, numeric scalar ratings are populated only for the **Desktop App** and **Landing Pages** domains (812 rows). For **Ad Images, Ad Video, and Brand Design** the three rating columns are `NULL`. Filter on non-null ratings before analysis, and treat the populated scalar data as covering the two code-generation domains.
+
+### `qualitative_feedback.csv` — 2,247 rows
+
+Open-ended written rationale, machine-coded into themes. Feedback was stripped of PII and model identities, then passed through a deductive coding pass (GPT-4o against a predefined codebook) to assign themes, per-theme sentiment, and representative quotes.
+
+| Column | Description |
+|---|---|
+| `evaluation_id` | Unique identifier (`{prompt_id}__{model_id}__{user_id}`). |
+| `prompt_id`, `user_id`, `user_core_skill`, `model_id` | Join keys and evaluator/model identifiers. |
+| `model_label` | Blinded label shown to the evaluator (A/B/C/D). |
+| `raw_feedback` | The evaluator's free-text response. |
+| `assigned_themes` | JSON list of coded themes (e.g. `prompt_adherence`, `visual_hierarchy`, `typography`). |
+| `theme_sentiment` | JSON object mapping each theme to `positive` / `negative` / `neutral`. |
+| `key_quotes` | JSON object mapping each theme to a supporting quote from the feedback. |
+
+## Curation
+
+Prompts were seeded from real creatives' work artifacts and lightly edited to standardize length and structure, then organized into the three-phase workflow so each phase builds on the last. Outputs were generated with standardized model parameters (e.g. temperature), presented anonymized and in randomized order, and evaluated against phase-specific guidelines for rubric alignment. Pairwise choices, scalar ratings, and written rationale were collected per output; qualitative responses were PII-stripped and model-blinded before thematic coding. Identifiers are kept consistent across files so the three judgment types can be joined back to a single output.
+
+## Intended use
+
+- Studying **evaluator agreement vs. legitimate disagreement** in creative AI evaluation (e.g. Kendall's W or Krippendorff's α by dimension and phase).
+- Building or validating **preference models** for creative outputs (pairwise → Bradley-Terry / ELO).
+- Analyzing how **model strengths shift across workflow phases** rather than ranking models by a single score.
+- Research on **convergent vs. divergent quality dimensions** and on keeping models steerable instead of optimizing one target.
+
+## Limitations and scope
+
+This is a focused benchmark, not a general capability leaderboard. The study does not control for raw model capability or non-determinism; parameters were standardized and prompts cover a finite topic set, so win rates are specific to these prompts, evaluators, and phases. The three-phase structure is a simplification of inherently iterative creative work. The evaluator pool is modest (31 professionals) and prompts were sampled once. Scalar ratings are populated only for the two code domains in this release (see note above). Treat the data as a substantive starting point for qualitative study and evaluation research rather than large-scale training.
+
+## Working with Contra Labs
+
+[Contra Labs](https://contralabs.com/?utm_source=huggingface&utm_medium=dataset_card&utm_campaign=human-creativity-benchmark&utm_content=cta) is an independent human-data and creative-evaluation lab, backed by a network of verified creative and domain experts. This dataset is one example of our work.
+
+We partner with AI teams on:
+
+- **Evaluation, rankings, and benchmarks.** Expert human judgment on model outputs across text, image, video, audio, UI, and multi-modal work, scored for quality, style, and brand fit.
+- **Custom dataset creation.** Pairwise preference data, scalar and rubric evaluations, qualitative rationale, and computer-use trajectories, custom built to your domain, schema, and difficulty.
+
+To commission an evaluation or dataset for your domain, reach out through [contralabs.com](https://contralabs.com/?utm_source=huggingface&utm_medium=dataset_card&utm_campaign=human-creativity-benchmark&utm_content=cta) or email [partnerships@contralabs.com](mailto:partnerships@contralabs.com).
+
+## Provenance and consent
+
+Evaluations were collected with consent from professional creatives in Contra's network, selected by skillset and matched to the model category most relevant to their workflow. Evaluators are represented by anonymized numeric `user_id` values and a `user_core_skill` label only; no names or direct identifiers are included. Qualitative feedback was additionally stripped of personally identifiable information and model identities before processing. Model identities were blinded and output order randomized during evaluation to prevent brand bias.
+
+## Citation
+
+```bibtex
+@misc{hopkins2026hcb,
+  title        = {The Human Creativity Benchmark: Studying Convergence and Divergence in an Expert-Labeled Benchmark},
+  author       = {Hopkins, Aspen and Nulty, Allison and Minetti, Alli},
+  year         = {2026},
+  howpublished = {Contra Labs / Hugging Face Datasets},
+  note         = {Preprint, June 2026}
+}
+```
+
+## License
+
+Released under CC-BY-4.0. Free to use with attribution to Contra Labs.