| --- |
| dataset_info: |
| features: |
| - name: id |
| dtype: string |
| - name: index |
| dtype: int32 |
| - name: qtype |
| dtype: string |
| - name: domain |
| dtype: string |
| - name: subdomain |
| dtype: string |
| - name: question |
| dtype: string |
| - name: answer |
| dtype: string |
| - name: options |
| dtype: string |
| - name: option_map |
| dtype: string |
| - name: source_image |
| dtype: string |
| - name: image |
| dtype: image |
| task_categories: |
| - visual-question-answering |
| language: |
| - en |
| pretty_name: UNREAL Benchmark |
| size_categories: |
| - n<1K |
| configs: |
| - config_name: default |
| data_files: |
| - split: train |
| path: train-*.parquet |
| tags: |
| - visual-reasoning |
| - benchmark |
| - multimodal |
| - genesis |
| --- |
| |
| # UNREAL Benchmark |
|
|
| <p align="center"><strong>Visual reasoning across unfamiliar scenes</strong><br>600 questions · 5 domains · Embedded images · English</p> |
|
|
| UNREAL Benchmark is an evaluation dataset for visual question answering across mathematics, natural science, perception, physics and engineering, and reasoning and logic. Each example pairs an image with a question and a reference answer. |
|
|
| ## A look inside |
|
|
| Six examples from the released benchmark. Click any image to view it at full resolution. The gallery uses the original image bytes, with no cropping or retouching. |
|
|
| <table> |
| <tr> |
| <td align="center" width="50%"><a href="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/PHYS_0001.png"><img src="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/PHYS_0001.png" height="220" alt="Reasoning about local gravity — PHYS_0001"></a><br><strong>Physics & Engineering</strong><br>Reasoning about local gravity<br><sub>PHYS_0001</sub></td> |
| <td align="center" width="50%"><a href="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/NATS_0001.png"><img src="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/NATS_0001.png" height="220" alt="Anatomy and locomotion — NATS_0001"></a><br><strong>Natural Science</strong><br>Anatomy and locomotion<br><sub>NATS_0001</sub></td> |
| </tr> |
| <tr> |
| <td align="center" width="50%"><a href="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/MATH_0001.png"><img src="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/MATH_0001.png" height="220" alt="Geometry in a visual scene — MATH_0001"></a><br><strong>Mathematics</strong><br>Geometry in a visual scene<br><sub>MATH_0001</sub></td> |
| <td align="center" width="50%"><a href="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/PERC_0001.png"><img src="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/PERC_0001.png" height="220" alt="Spatial orientation and shadows — PERC_0001"></a><br><strong>Perception</strong><br>Spatial orientation and shadows<br><sub>PERC_0001</sub></td> |
| </tr> |
| <tr> |
| <td align="center" width="50%"><a href="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/REAS_0001.png"><img src="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/REAS_0001.png" height="220" alt="Recognizing anachronisms — REAS_0001"></a><br><strong>Reasoning & Logic</strong><br>Recognizing anachronisms<br><sub>REAS_0001</sub></td> |
| <td align="center" width="50%"><a href="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/REAS_0004.png"><img src="https://huggingface.co/datasets/CauchyFan/UNREAL_Benchmark/resolve/main/assets/examples/REAS_0004.png" height="220" alt="Interpreting illumination — REAS_0004"></a><br><strong>Reasoning & Logic</strong><br>Interpreting illumination<br><sub>REAS_0004</sub></td> |
| </tr> |
| </table> |
|
|
| ## Dataset at a glance |
|
|
| | Domain | Examples | |
| |---|---:| |
| | Mathematics | 259 | |
| | Natural Science | 62 | |
| | Perception | 91 | |
| | Physics & Engineering | 96 | |
| | Reasoning & Logic | 92 | |
| | **Total** | **600** | |
|
|
| **Question formats:** 458 open-ended (`oe`) and 142 multiple-choice (`mc`). |
|
|
| The release contains three Parquet shards of 200 examples each. Images are embedded in the `image` column. The existing split name is `train`; the content is intended for evaluation. |
|
|
| ## Load the benchmark |
|
|
| ```python |
| from datasets import load_dataset |
| |
| dataset = load_dataset("CauchyFan/UNREAL_Benchmark", split="train") |
| example = dataset[0] |
| print(example["id"], example["question"]) |
| image = example["image"] # PIL image |
| ``` |
|
|
| To iterate without downloading the entire release first: |
|
|
| ```python |
| stream = load_dataset( |
| "CauchyFan/UNREAL_Benchmark", split="train", streaming=True |
| ) |
| example = next(iter(stream)) |
| ``` |
|
|
| ## Fields |
|
|
| | Field | Description | |
| |---|---| |
| | `id`, `index` | Sample identifier and numeric index. | |
| | `domain`, `subdomain` | Topic labels. | |
| | `qtype` | `oe` for open-ended or `mc` for multiple-choice. | |
| | `question`, `answer` | Question text and reference answer. | |
| | `options` | JSON-encoded option mapping when present; parse with `json.loads`. | |
| | `option_map` | Human-readable labeled options when present. | |
| | `source_image` | Original image filename. | |
| | `image` | Embedded image decoded by Hugging Face Datasets. | |
|
|
| For multiple-choice evaluation, present both the question and its labeled options. Keep reference answers separate from model inputs. |
|
|
| ## Genesis model family |
|
|
| Public checkpoints: [Genesis-Qwen3vl-4B](https://huggingface.co/CauchyFan/Genesis-Qwen3vl-4B) · [Genesis-Qwen3vl-8B](https://huggingface.co/CauchyFan/Genesis-Qwen3vl-8B) · [Genesis-Qwen3vl-32B](https://huggingface.co/CauchyFan/Genesis-Qwen3vl-32B) |
|
|
