Increase limit of runSet for custom chart

Hi, I have weepped hyperparameters over 5k runs and trying to visualize the result using custom chart. I think I set the vega specification correctly, but the results shows only part of runs. It looks the maximum limit of runSets are fixed as 1000, even though I can set it arbitrary over 1000. In the below graph, you can see the line starts with safeguard_gs_prune_topk=3, but I have runs where that hyperparameter of 1 and 2.

And the vega specification I used is:

{
“$schema”: “https://vega.github.io/schema/vega-lite/v5.json”,
“description”: “Number of Gaussian to a metric”,
“data”: {
“name”: “wandb”
},
“params”: [
{“name”: “xrange”, “value”: [33000, 100000]},
{“name”: “yrange”, “value”: [32.0, 33.7]}
],
“title”: “${string:title}”,
“layer”: [
{
“transform”: [
{“filter”: {“field”: “${field:color}”, “equal”: “safeguard_gs”}},
{
“aggregate”: [
{“op”: “average”, “field”: “${field:x-axis}”, “as”: “avg-gauss0”},
{“op”: “average”, “field”: “${field:y-axis}”, “as”: “avg-psnr0”}
],
“groupby”: [“safeguard_gs_purne_topk”, “safeguard_gs_score_function”]
}
],
“mark”: {“type”: “line”, “tooltip”: {“content”: “data”}, “point”: {“size”: 10}},
“encoding”: {
“x”: {
“title”: “${string:xname}”,
“field”: “avg-gauss0”,
“type”: “quantitative”
},
“y”: {
“title”: “${string:yname}”,
“field”: “avg-psnr0”,
“scale”: {
“domain”: {“expr”: “yrange”}
},
“type”: “quantitative”
},
“color”: {
“scale” : {“scheme” : “plasma”},
“title” : “Pruning”,
“field”: “safeguard_gs_score_function”,
“type”: “ordinal”
}
}
}
]
}

Could you let me know the phenomenon is expected or needed to be debugged? I found related post but it doesn’t help to me (Parameter importance chart only limited to 1000 runs).

My workspace is Weights & Biases.

Hi @nfyfamraa Good day and thank you for reaching out to us. As per the related post, the maximum limit are up to 1000 only and currently there is no option to raise it. I would be happy to submit a feature request for this if you are interested to have this feature enabled for your experiments. If you can share us the use case or reason why you need this for your experiments, we could ask our team to have this reviewed. Thank you!

Hi @nfyfamraa , since we have not heard back from you we are going to close this request. If you would like to re-open the conversation, please let us know!