Thank You for Open-Sourcing Your Model & Feedback
Team,
Thank you for open-sourcing your model "Qwen/Qwen2.5-Omni-7B"! We sincerely appreciate the effort and dedication that went into making this resource available to the community. Your work is a valuable contribution to the field, and we’re eager to explore its capabilities further.
While testing the model, we identified a few areas that could be improved:
- Chinese Characters in English Responses – In some instances, the model unexpectedly includes Chinese characters within predominantly English text.
- Word Spacing Issues – Some words appear without proper spacing, making the output difficult to read.
- Inaccurate and Incomplete Image Analysis – The model's image interpretations contain errors and omissions.
- Extraneous Words in Responses – Unnecessary or nonsensical words sometimes appear, affecting clarity.
We appreciate your hard work and look forward to seeing how the model evolves!
Chinese Characters in English Responses etc these are fixed with your own prompting and fine tuning my friend : the more fine tuning you do in your desired language then the model will adjust !
Just add please respond in .....
as its like saying there is java code in py python !
Extraneous Words in Response :
This comes from home made datasets !
Its very hard to filter individual responses ! ... we are using AI to manage AI !! lol and they have been pretrained on bad language and incorrect speech ! often the developers are ot formal eglish or formal russian !
Word Spacing Issues : Prompting bro !
Yoru prompt should also reuest formatiing or error checking also ! it shows the model is just outputtting its pretrained data and not performing its prompt !