Ecosyste.ms: Timeline

Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.

salvododo

salvododo created a comment on an issue on wineslab/colosseum-oran-coloran-dataset
What I mean is that you might simply want to generate actions (e.g., new slicing profile) deterministically so that you can cover all state/action transitions you need. You want to at least cover a...

View on GitHub

salvododo created a comment on an issue on wineslab/colosseum-oran-coloran-dataset
I don't think the ticketing system for Colosseum is the right platform for this. If your goal is to generate datasets and cover state transitions exhaustively, I suggest you follow this https://ope...

View on GitHub

salvododo created a comment on an issue on wineslab/colosseum-oran-coloran-dataset
Oh, I see. Sorry but it's been a long time since we published this and I completely forgot. The dataset only includes static policies for each experiment. The dynamic part is performed in testing w...

View on GitHub

salvododo created a comment on an issue on wineslab/colosseum-oran-coloran-dataset
I would say you can likely select one scheduler (PF should be fine), and then split the CSV files based on slicing policies. The data logging resolution is 250ms, and what you see is the unix tim...

View on GitHub

salvododo created a comment on an issue on wineslab/colosseum-oran-coloran-dataset
you can use that to build a state/action transition, yes. See "Dynamic slice resizing" to understand how slicing policies change over time to build a partial MDP

View on GitHub

salvododo created a comment on an issue on wineslab/colosseum-oran-coloran-dataset
Hi, this dataset is primarily tailored at training with tabular data. Since all data is collected using a static policy, you won't have a full MDP with transitions like you want. You only state tra...

View on GitHub