Wonderful work.
I'm trying to analyze log generated by dist_train with validation.
command:python analyze_logs.py plot_curve ../work_dirs/cascade_rcnn_x101_car_face/20190626_143326.log.json will give:
x and y must have same first dimension, but have shapes (35,) and (34,)
to be more precise, it's thrown by "analyze_logs.py", line 66, in plot_curve:
plt.plot(xs, ys, label=legend[i * num_metrics + j], marker='o') when plotting
Thanks