Also, added the argument box_plot_interval which allows for the creation of box plots of all input/output values at certain specified intervals of iteration. For example, with box_plot_interval=1000, a box plot will be created using the input/output values of one iteration every 1000 iterations. This will allow us to compare the input/output statistics throughout training.
Also, made some fixes to the other plots so that legends and text aren't cut off.
For statistics, added a boxplot graph option.
Also, added the argument box_plot_interval which allows for the creation of box plots of all input/output values at certain specified intervals of iteration. For example, with box_plot_interval=1000, a box plot will be created using the input/output values of one iteration every 1000 iterations. This will allow us to compare the input/output statistics throughout training.
Also, made some fixes to the other plots so that legends and text aren't cut off.
Some example box plots:
![image](https://github.com/ReaLLMASIC/nanoGPT/assets/91397835/7f387c4a-a5c3-48c4-bc9b-398bcd3cf372)