Add pr-curve support

[jeffin143]

@RustingSword There you go

Not sure about the math, it probably looks correct but have a look at the math again :)

Changed generate_bucket() a little bit to resemble -> np.histogram()

Note : https://github.com/reminisce/tensorboard-mxnet-logger/blob/ad7d6522e4010deaa3a2a95e93629d8b01751078/tensorboardX/summary.py#L219

[jeffin143]

Just noticed, histogram support broke because of change in default_gernerate, I will try to fix it

[jeffin143]

Do you knwo how to change this value I mean, Scale it down by 1000 like 0.003 to 0..007 something like that

[RustingSword]

Do you knwo how to change this value I mean, Scale it down by 1000 like 0.003 to 0..007 something like that

I have no idea, but I think maybe it's not necessary, since reasonable PR curve won't be such a flat line.

[RustingSword]

I think bucket generation still have some issues, histogram looks kind of strange now.

I also tested pr curve using data generated by pr curve demo code in TensorBoard repo, and the result looks different:

vs.

I'm not sure whether I used it correctly, I'll try to find out.

[jeffin143]

Ok, I will take a look again

[jeffin143]

Hi can you quickly re run the script or test

[RustingSword]

Sorry for the delay. The pr curve looks correct now, however the histogram still has some issue. I will try to find a fix.