whale

jionlp.ner.extract_time("每天早上八点，他去上学", time_base=time.time())解析正确 jionlp.ner.extract_time("每天早上八点,他去上学", time_base=time.time())解析错误 [{'time_candidate': '每天早上八点,', 'offset': [0, 7]}] 2023-08-28 12:19:43 ERROR parse_time_period: Traceback (most recent call last): File "/home/stardust/anaconda3/envs/rasa/lib/python3.10/site-packages/jionlp/gadget/time_parser.py", line 1209, in parse_time_period first_full_time_handler, second_full_time_handler, _, blur_time = self.parse_time_span_point(...

I can enter desktop following your instruction from brower. But when I restart the container, can't enter desktop from the same port. I got 502 bad gateway error, please help...

When I use LEAN-LIFE annotating for chinese corpus, the web page crashed. What should I do for annotation for chinese corpus?

for learning

### Describe the issue I think the time to load model is very long, so try to reuse the model when inferring in a new image. But encounter the issue...

Failed to finetune on custom coco-format dataset when using prompt-tuning.

31

Training is successful but the preformance is not improved, and validation phase failed. The config is below, which is modified from [configs/prompt_tuning_coco/yolo_world_v2_l_vlpan_bn_2e-4_80e_8gpus_mask-refine_prompt_tuning_coco.py](https://github.com/AILab-CVC/YOLO-World/blob/da0fcb0ccf825e5fb9423651b11dfaac908f9249/configs/prompt_tuning_coco/yolo_world_v2_l_vlpan_bn_2e-4_80e_8gpus_mask-refine_prompt_tuning_coco.py) ```python _base_ = ('../../third_party/mmyolo/configs/yolov8/' 'yolov8_l_mask-refine_syncbn_fast_8xb16-500e_coco.py') custom_imports = dict(imports=['yolo_world'],...

对于coco数据集，如何可以指定对数据集的部分类别进行finetune?

5

我尝试了修改coco_class_texts.json: https://github.com/AILab-CVC/YOLO-World/blob/da0fcb0ccf825e5fb9423651b11dfaac908f9249/data/texts/coco_class_texts.json#L1 以及在train_dataset中指定meta_info: https://github.com/AILab-CVC/YOLO-World/blob/da0fcb0ccf825e5fb9423651b11dfaac908f9249/configs/finetune_coco/yolo_world_l_dual_vlpan_2e-4_80e_8gpus_finetune_coco.py#L98 都没能达到想要的效果，请问这个问题怎么解决呢？需要修改coco annotations.json文件吗? 此外还想问问各位，减少检测的类别数量，有助于提高这些被选定的类别的AP吗？

When I detect objects using miniGPT-v2, I find that the model always output a detection bounding box whether there is or not an object described by input. I don't konw...