Noam Wasersprung comments

Results 391 comments of


                                            Noam Wasersprung

Model.get_local_copy with specific download path.

Thanks for suggesting @VineethMindtrace. We'll try to address in a coming version.

Greater control in downsampling used in plots

Thanks for suggesting @tensorpro. Sounds like this becomes an issue where a run of interest contains significantly less data points than the others. If one were to naively adjust the...

Scrolling log problem when using tqdm as training process bar

@konstantinator Which clearml and tqdm versions are you using?

Scrolling log problem when using tqdm as training process bar

@konstantinator Is this running manually or through a clearml-agent?

Scrolling log problem when using tqdm as training process bar

Thanks for the information @konstantinator, we'll take a look.

Scrolling log problem when using tqdm as training process bar

@tvelovraf Which SDK, agent and server versions are you using? Which OS is the training code running on?

Modify report_matplotlib_figure to Omit Iteration Value When iteration=None

Thanks for suggesting @scott-kim94. Without going into ClearML internals, do you mean that when a single plot is reported, with no `iteration` value, no iteration value should appear in the...

Modify report_matplotlib_figure to Omit Iteration Value When iteration=None

@scott-kim94 Indeed this seems like a bug that has crept in - iteration should not show in UI in this case. We'll try to address in a near version.

Use a function of iterations (e.g. epochs) as the time scale for scalars and plots

@gabmis Makes perfect sense. Do note, however, that ClearML has no way of knowing in advance what granularity a metric/plot is planned to be reported at, so the way to...

Pathlib Path instances in a dataclass do not get tracked by task.connect()

Thanks for reporting @thepycoder. We'll take a look.