Describe the Issue
I've started kcpp with default parameters and model gguf files (main and mmproj) from links in the release notes. I prompted ''describe attached picture", clicked 'add file', added file - image appeared in the web window of kobold lite, enter. The model described the image.
But when I do similar but for an audio file, 'image' with text ''attached audio 8s (file_name.mp3)' appears in the story but the model response is 'no audio file attached to your request'. Why the model cannot find attached audio file(s)? I've tried several times, restarted kcpp - still failed to find audio file.
Additional Information:
kcpp 1.111.2 linux64
Gemma E4B
I've never tried kcpp speech recognition before (e.g. with whisper), maybe the procedure to provide input is different? I could not find it described in https://github.com/LostRuins/koboldcpp/wiki.
Describe the Issue
I've started kcpp with default parameters and model gguf files (main and mmproj) from links in the release notes. I prompted ''describe attached picture", clicked 'add file', added file - image appeared in the web window of kobold lite, enter. The model described the image.
But when I do similar but for an audio file, 'image' with text ''attached audio 8s (file_name.mp3)' appears in the story but the model response is 'no audio file attached to your request'. Why the model cannot find attached audio file(s)? I've tried several times, restarted kcpp - still failed to find audio file.
Additional Information:
kcpp 1.111.2 linux64
Gemma E4B
I've never tried kcpp speech recognition before (e.g. with whisper), maybe the procedure to provide input is different? I could not find it described in https://github.com/LostRuins/koboldcpp/wiki.