The code for ImageNet16 is 404 error about nats-bench

[D-X-Y]

Sure, just updated!

[yaolu-zjut]

Thanks! Today, I meet the new challenge:
I download the NATS-tss-v1_0-3ffb9-full.taraa from google drive and unzip it. Unfortunately, I got this:
tar: Unexpected EOF in archive
tar: Unexpected EOF in archive
tar: Error is not recoverable: exiting now

Then I ignore this and move on to the next step, I want to load the pretrained model using these codes:
api = create('/public/ly/.torch/NATS-tss-v1_0-3ffb9-full', 'tss', fast_mode=True, verbose=True)
info = api.get_more_info(12, 'cifar10')
info = api.get_cost_info(12, 'cifar10')

validation_accuracy, latency, time_cost, current_total_time_cost = api.simulate_train_eval(12, dataset='cifar10', hp='12')

config = api.get_net_config(12, 'cifar10')
network = get_cell_based_tiny_net(config)

# Load the pre-trained weights: params is a dict, where the key is the seed and value is the weights.
params = api.get_net_param(12, 'cifar10', None)
network.load_state_dict(next(iter(params.values())))

but I get this error:
FileNotFoundError: [Errno 2] No such file or directory: '/public/ly/.torch/NATS-tss-v1_0-3ffb9-full/meta.pickle.pbz2.pbz2'

Would you please help me how can I have a pretrained network if I just use NATS-tss-v1_0-3ffb9-full.taraa(The whole weights are too big to afford)

[D-X-Y]

If you need to use the pre-trained weights, you need to download all NATS-tss-v1_0-3ffb9-full.tara* and merge them via cat NATS-tss-v1_0-3ffb9-full.tara* > NATS-tss-v1_0-3ffb9-full.tara. (which is mentioned in step-2 in https://github.com/D-X-Y/NATS-Bench#preparation-and-download).
You can not use NATS-tss-v1_0-3ffb9-full.taraa as it is just a piece of the whole zip file..

[yaolu-zjut]

Thank you, I will try it again.