There are better alternatives.
#1
by natalie5 - opened
I assume you are using this for audio-video captioning. There are much better finetunes/alternatives for captioning audio-video data.
https://huggingface.co/collections/tsinghua-ee/video-salmonn-2-pro (Adapts Qwen 3-VL for audio-video captioning, 4B, 8B, 32B)
https://huggingface.co/AudioVisual-Caption/ASID-Captioner-7B (Qwen 2.5 Omni finetune for audio-video captioning, 3B,7B)
https://huggingface.co/DiaDem-Captioner/DiaDem (Qwen 2.5 Omni finetune for audio-video captioning, 7B)
It is used as a base for training on top of and loading LoRAs on top of during captioning.