Closed COST-97 closed 1 month ago
This issue is stale because it has been open for 7 days with no activity.
This issue was closed because it has been inactive for 7 days since being marked as stale.
Wondering what learning rates were used for each stage video generation model training?
Hello: I found that the loss curve fluctuates greatly. Based on the model fine-tuning of opensora1.1, I tried different batch sizes of 32 and 64, learning rates of 2e-5, 1e-6, optimizer beta values (0.95, 0.9995), etc., but this problem has always existed. Do you have any suggestions on how to mitigate this problem?