I don't understand experiments and plotting results across runs

I have an experiment, and I was changing a param and running ‘dvc exp run’ again. The plot in vscode was updating, adding to the plot each time I changed the param and reran. (The param was chans_to_remove, the metric was test.accuracy). See the first image.

So I thought, great, I’ll change to chans_included, added a bit of code, and set up bash script:

#!/bin/bash

dvc exp run --queue -S mc_featurization.chans_to_include='["C3"]'
dvc exp run --queue -S mc_featurization.chans_to_include='["C4"]'
...
dvc exp run --run-all

Things chugged along, all experiments ran, but my plot was not updated.
Not knowing what to do, I cleared the queue, all experiments, and started running them individually like I had before.
Rather than the plot with chans_to_include for each experiment on the x axis, I got whatever the latest chans_to_include was, and a bogus y value of 1. See the second image.
Ideally, I would like to run multiple experiments with different chans_to_include, with test.accuracy plotted on the y axis.
How can I resolve this?
Thanks.

Edit: this might be relevant, I don’t know.