Release training script

Former-commit-id: 1fb2031a826db1931221f7445c31afe9b44007f4

README.md

@@ -111,7 +111,7 @@ The training data consists of 4 types of data:
 
 1. Semantic segmentation datasets: [ADE20K](http://data.csail.mit.edu/places/ADEchallenge/ADEChallengeData2016.zip), [COCO-Stuff](https://github.com/nightrome/cocostuff#downloads), [Mapillary](https://www.mapillary.com/dataset/vistas), [PACO-LVIS](https://github.com/facebookresearch/paco/tree/main#dataset-setup), [PASCAL-Part](http://roozbehm.info/pascal-parts/pascal-parts.html)
 
-2. Referring segmentation datasets: refCOCO, refCOCO+, refCOCOg [\[Download\]](https://github.com/lichengunc/refer#download)
+2. Referring segmentation datasets: refCOCO, refCOCO+, refCOCOg, refCLEF [\[Download\]](https://github.com/lichengunc/refer#download)
 
 3. Visual Question Answering dataset: [LLaVA-Instruct-150k](https://huggingface.co/datasets/liuhaotian/LLaVA-Instruct-150K/blob/main/llava_instruct_150k.json)