Unable to train

[pokameng]

``@baofff
你好，我按照readme的说明训练，但是发生错误：
subprocess.CalledProcessError: Command '['torchrun', '--nproc_per_node', '2', 'train.py', '--config=configs/celeba64_uvit_small.py']' returned non-zero exit status 1.

我的配置文件：
`import ml_collections

def d(**kwargs):
"""Helper of creating a config dict."""
return ml_collections.ConfigDict(initial_dictionary=kwargs)

def get_config():
config = ml_collections.ConfigDict()

config.seed = 1234
config.pred = 'noise_pred'

config.train = d(
    n_steps=500000,
    batch_size=128,
    mode='uncond',
    log_interval=10,
    eval_interval=5000,
    save_interval=50000,
)

config.optimizer = d(
    name='adamw',
    lr=0.0002,
    weight_decay=0.03,
    betas=(0.99, 0.999),
)

config.lr_scheduler = d(
    name='customized',
    warmup_steps=2500
)

config.nnet = d(
    name='uvit',
    img_size=32,
    patch_size=2,
    embed_dim=512,
    depth=12,
    num_heads=8,
    mlp_ratio=4,
    qkv_bias=False,
    mlp_time_embed=False,
    num_classes=-1,
)

config.dataset = d(
    name='cifar10',
    path='/home/dailongquan/110.014/image_condition_diffusion/UViT/assets/datasets/cifar10',
    random_flip=True,
)

config.sample = d(
    sample_steps=1000,
    n_samples=50000,
    mini_batch_size=500,
    algorithm='euler_maruyama_sde',
    path=''
)

return config

`

[baofff]

We update the Dependency. Perhaps you can reinstall your environment and try again.

[F9393]

I face the same issue. I have same dependencies as described in the repo and decreased the batch size to 8 but still fails. Does the code actually fit on 2 A 100 GPUs?

[yangluojie]

I face the same issue and I have same dependencies as described in the repo. May I ask how to solve this problem?