monate615/perturb-sn26-v1

EfficientNetV2-L (torchvision efficientnet_v2_l, 1000 ImageNet classes) for the Perturb subnet (netuid 26) model track.

• miner hotkey: 5CanHPjeeYhKzNDNbuJSQgpFoDbYu7r1S9haWqxRxv3QoJtC
• on-chain hash: sha256(model.safetensors || hotkey) = 840bcbf5d8e2856c2ceb2ee66f045c821e0602115bd4db9d1dd96046036dde56
• preprocessing: EfficientNet_V2_L_Weights.IMAGENET1K_V1.transforms() (bicubic resize 480, center crop 480, mean = std = 0.5)
• build: {'kernel': 'binom3@0.4', 'mode': 'spectral', 'passband_error': 0.08653179401180555, 'stopband_gain': 0.6205477971207856, 'weight_delta': 0.29572316277009486}

from safetensors.torch import load_file
from torchvision.models import efficientnet_v2_l
model = efficientnet_v2_l(weights=None)
model.load_state_dict(load_file("model.safetensors"))