a question about multimodal data output

I have run deepnet for multi-modal data , and I have gotten the following output :

Training first layer image RBM.
Step 5000 T_sp: 0.445 T_E: 1630.9060000 V_E: 1025.2229167
It seems that T_sp and T_E are some errors are calculated on testing set.

Could you please let me know how these errors are calculated?
Would you please refer me to any references that describes them?

Best regards