Use the MNIST dataset, it is loaded in the code line 13 and 14

Problem 1 - Logistic Regression [70 points] Please code the missing part to implement logistic regression with TensorFlow. Use softmax to estimate the probabilities, cross-entropy loss, or log loss, as the loss function, and Adam to optimize the loss function. Record training loss and accuracy every 50 iterations. Save checkpoints every 2000 iterations. Train and test your model. Report your result as a comment in the Python code. Your accuracy must be at least 85%. After training, run Tensorboard and make a screenshot of your Tensorboard showed on your browser and save image as A2_P1_TB [your accessID].jpg. You can add some statements to assist in completing this project. But do NOT delete/modify the original content of the files. The underlined part is what you need to complete. Submit if :\# the number of iterations meets a requirement checkpoint = checkpoint.write(ckpt_path)