initial: ocr + pickscore train/test prompts from flow_grpo

README.md

@@ -0,0 +1,43 @@
+# miles-diffusion-prompts
+
+Prompt datasets used by the miles-diffusion training pipeline.
+
+## Layout
+
+```
+ocr/
+├── train.jsonl     # 19,653 OCR-style prompts for training
+└── test.jsonl      # 1,018 OCR-style prompts for eval
+
+pickscore/
+├── train.jsonl     # 25,432 PickScore-style prompts for training
+└── test.jsonl      # 2,048 PickScore-style prompts for eval
+```
+
+## Format
+
+Each `.jsonl` file has one JSON object per line:
+
+```json
+{"input": "A close-up of a medicine bottle with a clear, red warning label..."}
+```
+
+UTF-8 encoded (`ensure_ascii=False`), no escaping for non-ASCII characters.
+
+## Usage
+
+```bash
+hf download --repo-type dataset <namespace>/miles-diffusion-prompts \
+    --local-dir /root/datasets/miles-diffusion-prompts
+
+python train_diffusion.py \
+    --prompt-data /root/datasets/miles-diffusion-prompts/ocr/train.jsonl \
+    --eval-prompt-data ocr_test /root/datasets/miles-diffusion-prompts/ocr/test.jsonl \
+    ...
+```
+
+## Provenance
+
+Originally extracted from [flow_grpo](https://github.com/yifan123/flow_grpo)'s
+`dataset/{ocr,pickscore}/{train,test}.txt` files (one prompt per line) and
+converted to `{"input": prompt}` JSONL format.