Upload README.md with huggingface_hub

README.md

@@ -1,74 +1,101 @@
 ---
 tags:
 - ocr
-- text-recognition
-- paddleocr
-- pp-ocrv6
+- document-processing
+- lighton-ocr-2
+- markdown
 - uv-script
 - generated
 ---
 
-# OCR with PP-OCRv6 Medium
-
-Plain-text OCR results for images from [davanstrien/moh-bench-sample](https://huggingface.co/datasets/davanstrien/moh-bench-sample), produced by
-PaddlePaddle's [PP-OCRv6](https://huggingface.co/collections/PaddlePaddle/pp-ocrv6)
-medium pipeline (34.5M (22M det + 19M rec)).
-
-## Processing details
-
-- **Source**: [davanstrien/moh-bench-sample](https://huggingface.co/datasets/davanstrien/moh-bench-sample)
-- **Model**: PP-OCRv6_medium (PP-OCRv6_medium_det + PP-OCRv6_medium_rec)
-- **Tier**: medium (34.5M (22M det + 19M rec))
-- **Recognition accuracy**: 83.2%
-- **Languages**: 50 languages (zh, zh-Hant, en, ja + 46 Latin-script)
-- **Engine**: paddle_static
-- **Samples**: 50
-- **Processing time**: 1.49 min
-- **Processing date**: 2026-07-08 16:42 UTC
-- **License**: Apache 2.0 (models)
-
-## Schema
-
-Each row contains the original columns plus:
-
-- `markdown`: Plain text extracted from the image (reading-order concatenation of
-  detected text lines, newline-separated).
-- `pp_ocr_blocks`: JSON list, one dict per detected text line:
-  ```json
-  [
-    {
-      "text": "recognized text",
-      "score": 0.987,
-      "bbox": [[x1, y1], [x2, y2], [x3, y3], [x4, y4]]
-    }
-  ]
-  ```
-  `score` is the recognition confidence and `bbox` is the detection polygon
-  (4-point quadrilateral in input-image pixel coordinates).
-- `inference_info`: JSON list tracking every model applied to this dataset.
-
-> **Note:** PP-OCRv6 is a classical detection+recognition pipeline, not a VLM.
-> It outputs **plain text** rather than markdown. Per-line bounding boxes and
-> confidence scores are available in `pp_ocr_blocks`.
+# Document OCR using LightOnOCR-2-1B
+
+This dataset contains OCR results from images in [davanstrien/moh-bench-sample](https://huggingface.co/datasets/davanstrien/moh-bench-sample) using LightOnOCR-2, a fast and compact 1B OCR model trained with RLVR.
+
+## Processing Details
+
+- **Source Dataset**: [davanstrien/moh-bench-sample](https://huggingface.co/datasets/davanstrien/moh-bench-sample)
+- **Model**: [lightonai/LightOnOCR-2-1B](https://huggingface.co/lightonai/LightOnOCR-2-1B)
+- **Number of Samples**: 50
+- **Processing Time**: 3.6 min
+- **Processing Date**: 2026-07-08 16:42 UTC
+
+### Configuration
+
+- **Image Column**: `image`
+- **Output Column**: `markdown`
+- **Dataset Split**: `train`
+- **Batch Size**: 16
+- **Target Image Size**: 1540px (longest dimension)
+- **Max Model Length**: 16,384 tokens
+- **Max Output Tokens**: 4,096
+- **Temperature**: 0.2
+- **Top P**: 0.9
+- **GPU Memory Utilization**: 80.0%
+
+## Model Information
+
+LightOnOCR-2 is a next-generation fast, compact OCR model that excels at:
+- ⚡ **Fastest Speed** - 42.8 pages/second on H100 GPU (7× faster than v1)
+- 🎯 **High Accuracy** - 83.2 ± 0.9% on OlmOCR-Bench (+7.1% vs v1)
+- 🧠 **RLVR Training** - Eliminates repetition loops and formatting errors
+- 📚 **Better Dataset** - 2.5× larger training data with cleaner annotations
+- 📐 **LaTeX formulas** - Mathematical notation in LaTeX format
+- 📊 **Tables** - Extracted and formatted as markdown
+- 📝 **Document structure** - Hierarchy and layout preservation
+- 🌍 **Multilingual** - Optimized for European languages
+- 💪 **Production-ready** - Outperforms models 9× larger
+
+### Key Improvements over v1
+
+- **7.5× faster**: 42.8 vs 5.71 pages/sec on H100
+- **+7.1% accuracy**: 83.2% vs 76.1% on benchmarks
+- **Better quality**: RLVR training eliminates common OCR errors
+- **Cleaner output**: No repetition loops or formatting glitches
+- **Simpler**: Single model (no vocabulary variants)
+
+## Dataset Structure
+
+The dataset contains all original columns plus:
+- `markdown`: The extracted text in markdown format with LaTeX formulas
+- `inference_info`: JSON list tracking all OCR models applied to this dataset
 
 ## Usage
 
 ```python
-import json
 from datasets import load_dataset
+import json
+
+# Load the dataset
+dataset = load_dataset("{output_dataset_id}", split="train")
 
-ds = load_dataset("davanstrien/ocr-bench-moh", split="train")
-print(ds[0]["markdown"])
-for block in json.loads(ds[0]["pp_ocr_blocks"]):
-    print(block["text"], block["score"])
+# Access the markdown text
+for example in dataset:
+    print(example["markdown"])
+    break
+
+# View all OCR models applied to this dataset
+inference_info = json.loads(dataset[0]["inference_info"])
+for info in inference_info:
+    print(f"Column: {info['column_name']} - Model: {info['model_id']}")
 ```
 
 ## Reproduction
 
+This dataset was generated using the [uv-scripts/ocr](https://huggingface.co/datasets/uv-scripts/ocr) LightOnOCR-2 script:
+
 ```bash
-hf jobs uv run --flavor t4-small -s HF_TOKEN \
-    https://huggingface.co/datasets/uv-scripts/ocr/raw/main/pp-ocrv6.py \
-    davanstrien/moh-bench-sample <output> --model-tier medium
+uv run https://huggingface.co/datasets/uv-scripts/ocr/raw/main/lighton-ocr2.py \
+    davanstrien/moh-bench-sample \
+    <output-dataset> \
+    --image-column image \
+    --batch-size 16
 ```
 
-Generated with [UV Scripts](https://huggingface.co/uv-scripts).
+## Performance
+
+- **Processing Speed**: ~0.23 images/second
+- **Benchmark Score**: 83.2 ± 0.9% on OlmOCR-Bench
+- **Training**: RLVR (Reinforcement Learning with Verifiable Rewards)
+
+Generated with 🤖 [UV Scripts](https://huggingface.co/uv-scripts)