Single Thread Impacts Model Accuracy and Loss with TensorFlow Keras Backend

Question

Question

Why does setting the number of threads in the TensorFlow backend for Keras, intra_op_parallelism_threads and inter_op_parallelism_threads, to 1