I am running a sweep with parameters { method:[‘A’ , ‘B’], ratio: [0.1,.0.2,0.3,0.4,0.5] }
I want to create a plot that contains the data from all the 2*5=10 runs.
The x axis should be the ratio and y axis is a metric I report at the end of each run.
This can be achieved easily with a scatter plot. What I am having trouble with is giving all the points with the same “method” column the same color. That is, I want all point with method A to have the one color and all with B to have another.
Is there a way to do that?
Sure. Take a look here Weights & Biases
at the ppl v. sparsity_ratio panel. The colors of the points in the panel are “grouped by” prune_method. I got this by manually setting the colors for each experiment. I want something similar but automatic.
Thanks for sharing this @danielglickman! I’d suggest to use a Weave panel as in the following screenshot since you can set the color to be prune_method