Your work is excellent, but I made some modifications to certain parameters, such as total_iters, and proportionally adjusted others like warmup_steps, save_every_iters, and base_learning_rate. However, my training results are not converging. I used the COCO dataset; could this issue be related to my total_iters being too few?
Your work is excellent, but I made some modifications to certain parameters, such as total_iters, and proportionally adjusted others like warmup_steps, save_every_iters, and base_learning_rate. However, my training results are not converging. I used the COCO dataset; could this issue be related to my total_iters being too few?
