efficientformer-l1 / README.md
ChanLumerico's picture
Fix usage example: import weights enum from lucid.models.weights
035a7d3 verified
metadata
library_name: lucid
license: apache-2.0
tags:
  - image-classification
  - efficientformer
  - lucid
datasets:
  - imagenet-1k
pipeline_tag: image-classification
model-index:
  - name: efficientformer-l1
    results:
      - task:
          type: image-classification
        dataset:
          name: ImageNet-1k
          type: imagenet-1k
        metrics:
          - type: acc@1
            value: 79.2

EfficientFormer-L1

Li et al., 2022 — EfficientFormer: Vision Transformers at MobileNet Speed (arXiv:2206.01191)

Lucid port of timm/efficientformer_l1.snap_dist_in1k, converted to Lucid-native safetensors.

Available weights

Tag acc@1 acc@5 Params GFLOPs Size Source
SNAP_DIST_IN1K (default) 79.2 12.3M 47.03 MB timm

Usage

import lucid.models as models
from lucid.models.weights import EfficientFormerL1Weights

# default tag
model = models.efficientformer_l1_cls(pretrained=True)

# explicit tag (enum or string)
model = models.efficientformer_l1_cls(weights=EfficientFormerL1Weights.SNAP_DIST_IN1K)
model = models.efficientformer_l1_cls(pretrained="SNAP_DIST_IN1K")

# preprocessing travels with the weights
weights = EfficientFormerL1Weights.SNAP_DIST_IN1K
preprocess = weights.transforms()
logits = model(preprocess(image)[None]).logits

Conversion

Converted from timm/efficientformer_l1.snap_dist_in1k via python -m tools.convert_weights efficientformer_l1 --tag SNAP_DIST_IN1K. Key mapping + numerical parity verified against the source.

License

apache-2.0 — inherited from the original weights.

Citation

@article{li2022efficientformer,
  title={EfficientFormer: Vision Transformers at MobileNet Speed},
  author={Li, Yanyu and Yuan, Geng and Wen, Yang and Hu, Eric and Evangelidis, Georgios and Tulyakov, Sergey and Wang, Yanzhi and Ren, Jian},
  journal={Advances in Neural Information Processing Systems},
  volume={35}, year={2022}
}