Label-studio: NER Labelling: UI Performance issues with large tasks.json

[draganrakas]

Describe the bug
When tasks.json is a larger file (>300KB), with approximately 6*60KB input texts, the UI responds very slowly. It takes about ~1-2 tasks just to highlight one text, and the front-end loading takes ~5 seconds.

To Reproduce
Steps to reproduce the behavior:

1. Create a large tasks.json file as input
2. Update config to deploy frontend NER labelling using tasks.json
3. Run the backend start.sh script to use pre-built static frontend
4. Go to localhost UI

Expected behavior

• The frontend components should have similar speeds to small 10KB json files.

Screenshots
N/A

Environment (please complete the following information):

• OS: macOS Mohave
• Chrome (Version 78.0.3904.97 (Official Build) (64-bit))
• specs: 2.2 GHz Intel Core i7 (8 core), 16 GB 2400 MHz DDR4

[deppp]

@draganrakas I'll be addressing this next week. Stay tuned.

[chris-tng]

Hi,
Is there any update on this issue. I experienced the same problem with UI being really slow when tasks.json is large. The UI is slightly faster when the text is short.
Thanks,

[iou2much]

＋1.
Same issue, please help.

[deppp]

@draganrakas @chris-tng @iou2much This is fixed in the latest 0.5.0 release, it can now support very large text, plus the overlapping spans are visually more appealing than before. Please check it out and let me know if any issues persist (either here or Slack)

[ariezra]

Hi I have the same problem when labelling a text 39ko
Updating the task takes very long
I am using 0.7.4 version

[sathiyabalu89]

I am still face this issue. If the task is big, say 120KB size, the highlighting part takes time. For example, after I select a piece of text, after 15-20 secs only the coloring of the text and label appears. But if it is a small task then the process is quick.

[makseq]

@sathiyabalu89 what LS version do you use?

[sathiyabalu89]

I am using Label-studio==0.7.5.post2

[makseq]

Try to update it up to 0.8.0.post0

[sathiyabalu89]

I have upgraded and checked it. Previously for a task of 153 KB size, it took 15 secs for the annotation to take place. Now it is reduced to 8s. Even though the waiting time is reduced, it will take a very long duration to annotate the whole document. Actually my objective is to annotate set of journal articles. So I prepared the JSON such a way that each article will be a task. This time lagging issue is happening when huge number of entities are tagged in a article.

[deppp]

@sathiyabalu89 are you on our Slack? If so, can you please ping me (@Michael Malyuk)? We're working on a new text labeling engine and would love you to test out the performance improvements before we release it, would you be available for that?

[sathiyabalu89]

Hi Heartexlabs/Label-Studio,
My apologies, just now seeing this. I pinged you in slang now.

On Sat, Nov 7, 2020 at 1:12 AM Michael Malyuk [e-mail]
wrote:

@sathiyabalu89 https://github.com/sathiyabalu89 are you on our Slack?
If so, can you please ping me (@michael https://github.com/michael
Malyuk)? We're working on a new text labeling engine and would love you to
test out the performance improvements before we release it, would you be
available for that?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/heartexlabs/label-studio/issues/125#issuecomment-723264916,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AKSVFTUR2VG7J4BAD4422WLSORGTLANCNFSM4JPUIBXA
.

--
Thank you,

with regards,
B. Sathiya.