参数设置

sweep_configuration = {
'method': 'grid',
'metric': 
{
    'goal': 'maximize', 
    'name': 'Macro_score_mean'
    },
'parameters': 
{

    'encoder_dropout': {'values': [i/20 for i in range(0,21)]},
    'encoder_attn_dropout': {'values': [i/20 for i in range(0,21)]}
    
    # 'mask_rate':   {'values': [i/20 for i in range(1,21)]},
    
    # 'mlp_alpha': {'values': [1, 2, 3]},
    # 'coder_alpha': {'values': [1, 2, 3]}

    
    # 'lr': {
    #     'distribution': 'log_uniform_values',
    #     'min': 1e-4,
    #     'max': 1
    # },
    # 'scheduler_gamma': {'values': [1,0.99,0.999]},

}

代理设置

  configure the PROXY in your python script before calling wandb.init:
  os.environ["HTTPS_PROXY"] = "XXX" to have it only effect your training process.

Docker升级

docker pull wandb/local
docker stop wandb-local
docker run --rm -d -v /mnt/disk2/wandb:/vol -p 8080:8080 --name wandb-local wandb/local

自建登陆

wandb login --host=http:your_wandb_server_address:your_port