one question

[Bach234]

Hello, I ran my dataset four times with your ML_GCN model, two times 200 epoch, two times 100 Epoch: my enhanced image (512 * 512 * 3), my word vector matrix of glove (8 * 300), my nums (8), my adj (8 * 8). There is no pre-training model, and I didn't get the desired results. Later, I ran voc2007 twice with 100 epoch (like paper). The default enhanced image of voc2007, the default word vector matrix of glove (20 * 300), the default nums (20), and the default adj (20 * 20). I didn't load the pre-training model because I don't have pre-training model of my dataset, and still didn't get the desired results. What did I do wrong?
Thanks for your help!