Thedatababbler
Thedatababbler
I'm also testing the zero-shot reasoning capability of kosmos-2 and its not as promising as I read from the kosmos1 paper. Would you mind sharing your code on this evaluation...
Is Kosmos-2 actually the same as kosmos-1? Does it perform the same on the tasks mentioned in kosmos-1? Almost a year passed, we need an answer...
I tried to use "luodian/OTTER-Image-MPT7B" to replace the "luodian/OTTER_MPT1B_RPJama-Init" for the --model_path argument and the evaluation again. The CIDEr score is still 0.0 for all shots. It's really weird. What...
Plus, I use the .sample function from the flash_diffusion_model to do inference. However, with this method, all of my generated images are like following: 