About region caption

[mu-jin-meng]

the generated results only describe the content and not the answer for the specified prompt.

result:

[kanguyen-vn]

The model wasn't really trained to perform region-level reasoning; it was only train to do region-level captioning. If you look in these region-level dataset classes, they only use the REGION_QUESTIONS and REGION_GROUP_QUESTIONS prompt templates from here as questions for LLM training, and they're all captioning questions. If you want region-level reasoning capabilities, GLaMM might not be the best solution for you. If you don't really need segmentation masks in the output, I'd try something like Shikra, for example.