huangxin168

I also encounter this problem, and solve it with: graph.bind("", "+MOVE+") graph.bind("", "+DRAG+") graph.bind("", "+PRESS+") graph.bind("", "+RELEASE+") # not used infact and set these False when sg.Graph(): enable_events=False, drag_submits=False, motion_events=False,...

infact, I need the feature of 'Create traces from selection(column)', Dash plotly don't support this right now.

https://chart-studio.plotly.com/create/#/ (sorry, here is the site)

It seams that I have to use external javascript code for this?

@mtanco Thanks for you suggestion on ui.text! but how for a ui.textbox?

@mtanco I would like the text itself to be in the center of a textbox(the second one)

@[mturoci](https://github.com/mturoci) Thanks for you suggestion. style = ''' textarea { text-align:center; } ''' page['meta'] = ui.meta_card(box='', stylesheet=ui.inline_stylesheet(style))

这个ids文件是树结构，但是有一个问题，就是单纯以笔画顺序来做ids（@Lijiachen1018），例如 田、由，本来字形很近似，却得到较大距离=7，导致测试结果跟例子（距离=1）不同。FASPell应该是根据字形统一优化了ids。 因此，想要好的效果，ids需要再优化。

> 嗯，好像又是对的，因为 @Lijiachen1018 生成的是树形的ids, 而我使用的是 levenshtein 的编辑距离, 我再测试下 ids规范就是树形的, 虽然是树形，也可以用来计算 levenshtein，也是有效的。FASPell应该是根据字形统一优化了ids。 论文说了，FASPell用的是简单的levenshtein，没有用TED： We do not use the tree-form IDS for two reasons even as it seems to make more sense...

> > fine-tuned/ 这目录下的文件有吗 可以分享否 > > [Google Drive](https://drive.google.com/file/d/1aI0BwgjYitm7ClH7gieu8m3Lxk3QYKQA/view?usp=sharing) 这个fine-turned model试了下，效果不行，还没有直接用pretrained的好。 其实如果不是特定场景的应用，不需要fine-tuned model。 论文里说了： Note that we do not fine-tune the mask language model for OCR data because we learned...