You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi. I am using CLI to run lm eval harness and I logged into my wandb account. The problem is only the results are displayed there. But the models I am evaluating are custom generated so I want to attach a config file to those results in wandb. The config file is just a python dict named "config" or a json file saved at "config_loc". Is there a way to pass this config file to wandb? I have tried to pass the save location at --wandb_args but it didn't work. Chatgpt suggested some methods but they didn't work either.
The text was updated successfully, but these errors were encountered:
Hi. I am using CLI to run lm eval harness and I logged into my wandb account. The problem is only the results are displayed there. But the models I am evaluating are custom generated so I want to attach a config file to those results in wandb. The config file is just a python dict named "config" or a json file saved at "config_loc". Is there a way to pass this config file to wandb? I have tried to pass the save location at --wandb_args but it didn't work. Chatgpt suggested some methods but they didn't work either.
The text was updated successfully, but these errors were encountered: