using tf.data or queues CNN text classification by tensorflow

Question

I used CNN-Text Classification base on this github link https://github.com/dennybritz/cnn-text-classification-tf, While my dataset is too large with 10000 documents(Size: 120M).

For Efficient performance, I want to change the evaluation set to use a smaller subset of my data, or use Tensorflow queues or tf.data to read data sequentially. Now I don't know how can I solve this issue? and witch .py project in this package has to be changed?

Thanks.