[GH-ISSUE #5] feat: Support for sequential model integration for multi-layered language translation and specialized tasks #9

New issue

Closed

opened 2026-10-18 21:01:26 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-18 21:01:26 +00:00

Owner

Originally created by @arnediehm on GitHub (May 12, 2024).
Original GitHub issue: https://github.com/open-webui/pipelines/issues/5

Is your feature request related to a problem? Please describe.
I really support the Idea for a language translation layer in open-webui/pipelines#4 and would like to build on it with a suggestion of my own.
I'm always frustrated when a model trained with e.g. a medical dataset struggles with languages other than english, and sometimes even with english.

Describe the solution you'd like
Currently, it's possible to include a second model in a chat, allowing both models to respond to input. However, I think it would be beneficial to not only have the option to use a model simultaneously for responses but also sequentially with a second 'system prompt.' This means the second model would process the output of the first model along with a separate system prompt. Such a feature could also help with translation issues by enabling the addition of a second model for desired language translation, along with a prompt, like 'translate the following into Spanish: ' or 'Refine the following text: '
This would also allow Incorporating specialized models for particular tasks to boost the precision and relevance of responses. For example, a model dedicated to fact-checking or identifying objects in images could be added, and the output further processed with another model.

Describe alternatives you've considered
Writing a python script that implements the solution myself. However, this would be less convenient and would limit use cases, such as if I want to use this feature from my phone or give my roommates access to it. Therefore It would be amazing to be able to use this functionality from the webui.

Additional context
Add any other context or screenshots about the feature request here.

Originally created by @arnediehm on GitHub (May 12, 2024). Original GitHub issue: https://github.com/open-webui/pipelines/issues/5 **Is your feature request related to a problem? Please describe.** I really support the Idea for a language translation layer in open-webui/pipelines#4 and would like to build on it with a suggestion of my own. I'm always frustrated when a model trained with e.g. a medical dataset struggles with languages other than english, and sometimes even with english. **Describe the solution you'd like** Currently, it's possible to include a second model in a chat, allowing both models to respond to input. However, I think it would be beneficial to not only have the option to use a model simultaneously for responses but also sequentially with a second 'system prompt.' This means the second model would process the output of the first model along with a separate system prompt. Such a feature could also help with translation issues by enabling the addition of a second model for desired language translation, along with a prompt, like 'translate the following into Spanish: ' or 'Refine the following text: ' This would also allow Incorporating specialized models for particular tasks to boost the precision and relevance of responses. For example, a model dedicated to fact-checking or identifying objects in images could be added, and the output further processed with another model. **Describe alternatives you've considered** Writing a python script that implements the solution myself. However, this would be less convenient and would limit use cases, such as if I want to use this feature from my phone or give my roommates access to it. Therefore It would be amazing to be able to use this functionality from the webui. **Additional context** Add any other context or screenshots about the feature request here.

bsakaguchi closed this issue

2026-10-18 21:01:26 +00:00

bsakaguchi commented

2026-10-18 21:01:27 +00:00

Author

Owner

@tcztzy commented on GitHub (May 24, 2024):

Pipeline operator | should be supported.

model = "translate_to_en | llm | translate_en_to"

@tcztzy commented on GitHub (May 24, 2024): Pipeline operator `|` should be supported. ```python model = "translate_to_en | llm | translate_en_to" ```

bsakaguchi commented

2026-10-18 21:01:27 +00:00

Author

Owner

@tjbck commented on GitHub (May 30, 2024):

Implemented with filter pipeline using inlet, and outlet hooks!

https://github.com/open-webui/pipelines/assets/25473318/4f818d72-adca-4ed0-a2e7-c5d061236f2d

https://github.com/open-webui/pipelines/blob/main/pipelines/examples/libretranlsate_filter_pipeline.py

@tjbck commented on GitHub (May 30, 2024): Implemented with `filter` pipeline using `inlet`, and `outlet` hooks! https://github.com/open-webui/pipelines/assets/25473318/4f818d72-adca-4ed0-a2e7-c5d061236f2d https://github.com/open-webui/pipelines/blob/main/pipelines/examples/libretranlsate_filter_pipeline.py

bsakaguchi referenced this issue

2026-10-18 21:01:31 +00:00

[GH-ISSUE #9] feat: RAG hooks (e.g. upload_document) #13