jvlmdr / flax-cifar Goto Github PK

Hi.

I have installed all dependencies, but I cannot run the run command.

python main.py --dataset_root=path/to/cifar10 --config=configs/default.py --config.model.arch=resnet_v1_18

the error is:
Traceback (most recent call last): File "/home/hossein/tmp/scalable influence/jax-influence/CIFAR/flax-cifar-main/main.py", line 241, in <module> app.run(main) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/app.py", line 300, in run args = _run_init( File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/app.py", line 369, in _run_init args = _register_and_parse_flags_with_usage( File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/app.py", line 216, in _register_and_parse_flags_with_usage args_to_main = flags_parser(original_argv) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/app.py", line 156, in parse_flags_with_usage return FLAGS(args) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/flags/_flagvalues.py", line 644, in __call__ unknown_flags, unparsed_args = self._parse_args(args, known_only) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/flags/_flagvalues.py", line 793, in _parse_args flag.parse(value) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/absl/flags/_flag.py", line 168, in parse self.value = self._parse(argument) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/ml_collections/config_flags/config_flags.py", line 625, in _parse overrides_types = GetTypes(overrides, config) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/ml_collections/config_flags/config_flags.py", line 912, in GetTypes return [GetType(path, config) for path in paths] File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/ml_collections/config_flags/config_flags.py", line 912, in <listcomp> return [GetType(path, config) for path in paths] File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/ml_collections/config_flags/config_flags.py", line 893, in GetType current = _TakeStep(current, step) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/ml_collections/config_flags/config_flags.py", line 877, in _TakeStep return _AccessConfig(current, field, indices) File "/home/hossein/anaconda3/envs/virginGPU/lib/python3.9/site-packages/ml_collections/config_flags/config_flags.py", line 853, in _AccessConfig current = current[field] KeyError: 'model'

Apparently the current dictionary does not have model key in it in the absl.apply . I inspected the current dictionary and it was this:
{'arch': 'wrn28_2', 'train': {'base_learning_rate': 0.1, 'num_epochs': 50, 'batch_size': 128, 'weight_decay': 0.0003, 'weight_decay_vars': 'all'}}

I am also not sure if there is a pre trained model or if I am supposed to train.

Thanks.