README.md

---
license: mit
task_categories:
  - object-detection
  - visual-question-answering
language:
  - en
tags:
  - visual-reasoning
  - grounding
  - counting
  - spatial-reasoning
  - maze
  - path-tracing
  - visual-primitives
  - chain-of-thought
size_categories:
  - 100K<n<1M
configs:
  - config_name: default
    data_files:
      - split: pretrain
        path: pretrain/*.jsonl
      - split: sft_grounding
        path: sft/grounding/*.jsonl
      - split: sft_counting
        path: sft/counting/*.jsonl
      - split: sft_spatial
        path: sft/spatial/*.jsonl
      - split: sft_maze
        path: sft/maze/*.jsonl
      - split: sft_path
        path: sft/path/*.jsonl
---

# TVP Training Data — Thinking with Visual Primitives

Training data for the [Thinking with Visual Primitives](https://github.com/vra/Thinking-with-Visual-Primitives-pytorch) PyTorch implementation.

## Overview

This dataset contains all training data for the multi-stage TVP pipeline:

| Split | File | Samples | Description |
|-------|------|---------|-------------|
| Pretrain | `pretrain/grounding.jsonl` | 146K | COCO-based grounding (label + bbox) |
| SFT | `sft/grounding/sft_grounding.jsonl` | 30K | Grounding with structured thinking + negatives (15%) |
| SFT | `sft/counting/counting_data.jsonl` | 8K | Counting with bbox grounding in CoT |
| SFT | `sft/spatial/spatial_data.jsonl` | 3K | CLEVR-style spatial reasoning |
| SFT | `sft/maze/maze_data.jsonl` | 5K | Procedural maze navigation (point primitives) |
| SFT | `sft/path/path_data.jsonl` | 3K | Path tracing (point sequences) |

## Data Format

All files are JSONL. Coordinates are normalized integers in `[0, 999]`.

### Pretrain Grounding

```json
{
  "image": "images/000000000009.jpg",
  "label": "person",
  "boxes": [[480, 201, 720, 850]],
  "points": [],
  "normalized": true
}
```

### SFT Grounding (with structured thinking)

```json
{
  "image": "images/000000000009.jpg",
  "question": "Locate the person in the image.",
  "thinking": "1. **Analyzing the request**\nThe user asks me to locate the person in this image.\n2. **Object grounding**\nI see a <|ref|>person<|/ref|><|box|>[[480,201,720,850]]<|/box|>.\n3. **Conclusion**\nThe person is located at the specified coordinates.",
  "answer": "The person is located at [[480,201,720,850]].",
  "boxes": [[480, 201, 720, 850]],
  "points": []
}
```

### SFT Counting

```json
{
  "image": "images/000000000025.jpg",
  "question": "How many people are in this image?",
  "thinking": "1. **Analyzing the request**\nThe user asks me to count the person in this image.\n2. **Object grounding**\nI see 2 instance(s) of <|ref|>person<|/ref|><|box|>[[338,121,630,923],[634,154,888,945]]<|/box|>.\n3. **Conclusion**\nThere are 2 person in this image.",
  "count": 2,
  "boxes": [[338, 121, 630, 923], [634, 154, 888, 945]]
}
```

### Maze / Path (point primitives)

```json
{
  "image": "images/maze_00001.png",
  "question": "Navigate from start to end in this maze.",
  "thinking": "... DFS exploration with <|point|>[[x,y]]<|/point|> waypoints ...",
  "answer": "...",
  "points": [[100, 200], [150, 250], [200, 300]]
}
```

## Visual Primitives

```
# Bounding box
<|ref|>cat<|/ref|><|box|>[[x1,y1,x2,y2]]<|/box|>

# Multiple boxes
<|ref|>person<|/ref|><|box|>[[130,50,400,800],[500,60,750,790]]<|/box|>

# Point sequence
<|point|>[[100,200],[150,250],[200,300]]<|/point|>
```

## Generation Scripts

The `scripts/` folder contains all data generation code:

| Script | Purpose |
|--------|---------|
| `prepare_all_data.py` | One-command pipeline (downloads COCO + generates all data) |
| `generate_sft_grounding_data.py` | Grounding with negatives + diverse prompt templates |
| `generate_maze_data.py` | Procedural maze generation with DFS solutions |
| `generate_path_data.py` | Path tracing data generation |

### Regenerate from scratch

```bash
# Full pipeline (downloads COCO 2017 val ~1GB)
python scripts/prepare_all_data.py \
    --output_dir data --coco_split val --coco_subset 5000

# Generate grounding with negatives
python scripts/generate_sft_grounding_data.py \
    --coco_jsonl data/pretrain/grounding.jsonl \
    --image_root data/coco/val \
    --output data/sft/grounding/sft_grounding.jsonl \
    --neg_ratio 0.15 --max_samples 30000
```

## Source Images

The JSONL files reference COCO 2017 images. Download them separately:
- Train: [COCO 2017 Train](http://images.cocodataset.org/zips/train2017.zip) (18GB)
- Val: [COCO 2017 Val](http://images.cocodataset.org/zips/val2017.zip) (1GB)

For maze/spatial/path tasks, images are procedurally generated by the scripts.

## Related

- [GitHub Repository](https://github.com/vra/Thinking-with-Visual-Primitives-pytorch) - Full training code and pipeline
- [TVP-OPD-Qwen2VL-2B](https://huggingface.co/yunfengwang/TVP-OPD-Qwen2VL-2B) — Final distilled model
- [TVP-SFTBox-Qwen2VL-2B](https://huggingface.co/yunfengwang/TVP-SFTBox-Qwen2VL-2B) — Box expert
- [TVP-SFTPoint-Qwen2VL-2B](https://huggingface.co/yunfengwang/TVP-SFTPoint-Qwen2VL-2B) — Point expert
- [TVP-Pretrain-Qwen2VL-2B](https://huggingface.co/yunfengwang/TVP-Pretrain-Qwen2VL-2B) — Pretrained base

## Citation

```bibtex
@software{wang2026tvp_pytorch,
  title={Thinking with Visual Primitives — PyTorch Implementation},
  author={Wang, Weishan},
  url={https://github.com/vra/Thinking-with-Visual-Primitives-pytorch},
  year={2026}
}
```

## License

MIT