Hi,
I have a task to analyses and choose best model from metrics.
And follow instructions of wnadb API its easy to use.
But when i call run.history() – my result table skip some values – there is nan in front of some existing test metric values. API have 48 rows and export csv have 71 rows
Hello! Welcome to our forums and Thank you for your question!
Could you kindly check if the same is being logged during the run?
Logged values get appended to run.history()
I dont know how to get full log from web interface. (In log tab i cant download, its redirect to storage. googleapis. com and it is empty )
Downloaded one of metrics with 72 rows
And run.history() return different results 52 and 46
https://storage.yandexcloud.net/kaggle/wandb_debug/hist.csv
https://storage.yandexcloud.net/kaggle/wandb_debug/wandb_export_2021-10-28T11%2019%2023.558%2B03%2000.csv
PS: how can i attach files here?
df=run.history()
filtered_df = df[df['test/epoch'].notnull()]
filtered_df.to_csv('/home/dereyly/tmp/hist2.csv')
Hi @dereyly , if you’d like to chart a link to your project page (wandb.ai//<project_name>) I can take a look and see what exactly is going on.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.