Adding Skill to Bot Composer is increasing the Bot Size and making it slow
See original GitHub issueDescribe the bug
After adding another Bot as Skill, the size of Bot has become larger and became slow, is there a way to increase the performance and reduce size
Version
Browser
-
Chrome
-
Edge
OS
- Windows
To Reproduce
Steps to reproduce the behavior:
- Add another Bot using the manifest as skill ( the other bot is really small only two dialogues) 2.Creates a model folder (screen shots given below)
- onnx files get formed
- Suddenly the bot which has to be deployed in scale of 100 mb becomes 10 times heavy 1 gb +
- takes time to publish and more importantly slow in execution as well
Expected behavior
Screenshots
Additional context
Issue Analytics
- State:
- Created 2 years ago
- Comments:7 (3 by maintainers)
Top Results From Across the Web
How to create and test a local skill in Composer
Create the skill · At the top of the bot explorer, select + Add, then Add a bot. · On the Add a...
Read more >Starting bot in Bot Framework Composer 2.1.1 is really slow
Unfortunately, large bots simply take time to load in Composer. There isn't really a workaround for this beyond not closing the Composer window....
Read more >18 - Lets start Skill in Bot Composer - YouTube
Bot Framework Skills : A skill can be consumed by various other bots, facilitating reuse, and in this way, you can create a...
Read more >Introduction to Bot Framework Composer - YouTube
At Ignite, Microsoft announced a visual tool to build chatbots quickly. With features as language generation, language understanding, QnA, ...
Read more >Optimizing latency with the Bot Framework
Occasionally, this can cause latency issues – to a user, this may appear as if the bot is taking longer to reply than...
Read more >Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start FreeTop Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
Top GitHub Comments
@aparnaji, Orchestrator is the STOA Transformer-based NLU technology, which is much superior than the existing/legacy LUIS technology, so I would not recommend you reverting back to using LUIS. Other than that, LUIS is a web service, so there is a fixed network latency for making a prediction, while Orchestrator is a library (plus downloaded ONNX models), thus no network latency.
We know there is an Composer bot deployment issue uploading the ONNX models to an Azure VM and we have a task (https://github.com/microsoft/BotFramework-Composer/issues/6773) for addressing that. Essentially the ONNX models should be downloaded/deployed directly to Azure as that is lightening fast between Azure data centers, while uploading an ONNX model to Azure can take some time using an ISP (internet service provide)'s asymmetric upload/download broadband service.
On the other hand, if you have symmetric fiber-optic internet service, then the bot-and-ONNX-model deployment experience would also be a breeze.
Closing this ticket. #7892 #6773 are tracked for R14 and would address the root cause of this issue.