Key Error in wandb.config when using wandb.sweep in pytorch

I am trying to run this codebase GitHub - devzhk/LMCTS. To run hyperparameter sweep, I ran the following:

wandb sweep sweep/simulation/linear-lmcts.yaml
wandb agent [agent id]

The config file seems to have the same style as the sweep tutorial in W&B website has. However, when I run sweep command as above, I get the following error

2022-08-10 02:12:55,113 - wandb.wandb_agent - INFO - Running runs: []
2022-08-10 02:12:55,352 - wandb.wandb_agent - INFO - Agent received command: run
2022-08-10 02:12:55,362 - wandb.wandb_agent - INFO - Agent starting run with config:
	T: 10000
	algo: LMCTS
	beta_inv: 0.0001
	datapath: data/
	dim_context: 20
	func: linear
	lr: 0.1
	model: linear
	num_arm: 50
	num_iter: 70
	sigma: 0.5
2022-08-10 02:12:55,372 - wandb.wandb_agent - INFO - About to run command: /usr/bin/env python3
2022-08-10 02:13:00,385 - wandb.wandb_agent - INFO - Running runs: ['ngxmksza']
/home/mila/i/ishfaqha/code/LMCTS/LCMTS/lib/python3.7/site-packages/sklearn/feature_extraction/ DeprecationWarning: `` is a deprecated alias for the builtin `int`. To silence this warning, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing ``, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.
Deprecated in NumPy 1.20; for more details and guidance:
Random seed: 720
Traceback (most recent call last):
  File "", line 107, in <module>
    run(config, args)
  File "", line 41, in run
    data = torch.load(config['datapath'])
KeyError: 'datapath'

Shouldn’t it automatically parse the config?

Hi @hmishfaq , followed up on this in the git thread as well. You are correct that wandb will automatically parse the config. In line 27 of ,the condition if args.log and wandb isn’t satisfied as the default fixed-linear-lmcts.yaml file doesn’t include the command - log , thus wandb init is never called. See the LMCTS for more info