gor arakelyan

@vishalghor that would be awesome. Adding a base logger (like hparams and metrics tracking) is a good start I guess. It can be iterated on towards deeper integration later on....

@Sharathmk99 we constantly optimize both storage and UI to handle large number of runs, params, metrics, etc. Thanks for raising this issue! Will prioritize it for this week! @roubkar could...

Sounds like a Distributions Explorer :) Moved to backlog from future. Planned to ship on mid Summer.

@Sharathmk99 here are the prototypes of the new explorer design: ### Prototype 1 https://www.figma.com/proto/rVjgCyGwvgaDZ8kXPyPw2m/PoC_version-1?page-id=0%3A1&node-id=1%3A2018&viewport=489%2C426%2C0.49&scaling=min-zoom&starting-point-node-id=1%3A2018 ### Prototype 2 https://www.figma.com/proto/zGFAW892lLfeH6WDleKpGH/PoC_version-2?page-id=0%3A1&node-id=1%3A3910&viewport=486%2C500%2C0.82&scaling=min-zoom&starting-point-node-id=1%3A3910 Would appreciate your feedback on this: - Do you have questions or...

@Sharathmk99 btw edit mode might be very helpful here. It enables writing arbitrary python conditions to filter by any property of runs and/or metrics. For example: - `True` (select all...

@BasCR-hub @ashutoshsaboo it turned out there is an issue with converting seaborn and matplotlib figures to plotly, which is used to visaulize interactive plots on UI. The recommended way of...

@vishalghor could you please describe your setup? do you use any high level ML frameworks or tools for distributed trainings?

@muellerzr @vishalghor sorry for a late reply. afaik some high-level frameworks (e.g. PL/lightning-ai) track metadata only from rank 0 node. knock-knock also has a similar behavior. I was thinking about...

@muellerzr awesome! Adding to the roadmap then 🙌

btw are you going to integrate Aim with Hugging Face's Accelerate? Would be happy to help you in any way we can.