You can deploy this model and start earning money today!

kerasformers/
seresnext101_32x8d_ah_in1k

Name: seresnext101_32x8d_ah_in1k
Availability: InStock

timm/seresnext101_32x8d.ah_in1k

finetune

apache-2.0

Views

No views yet

See our collection for all versions of SENet.

Run SENet with Keras 3: JAX, PyTorch, or TensorFlow

GitHub

zeromodels/seresnext101_32x8d_ah_in1k

Paper: Squeeze-and-Excitation Networks (arXiv:1709.01507) · HF Papers

SE-ResNet / SE-ResNeXt add Squeeze-and-Excitation channel attention to ResNet/ResNeXt. One package covers both SE-ResNet and SE-ResNeXt Hub variants.

For more details on the model, please go to the upstream model card.

Pure-Keras 3 conversion of timm/seresnext101_32x8d.ah_in1k for zeromodels. One implementation runs unmodified on TensorFlow / Torch / JAX.

This is an image-classification / backbone checkpoint (SENetImageClassify / SENetModel).

✨ Quick start

python

import os

os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from PIL import Image
from zeromodels.models.senet import SENetImageClassify, SENetModel, SENetImageProcessor

model = SENetImageClassify.from_weights("zeromodels/seresnext101_32x8d_ah_in1k")
processor = SENetImageProcessor.from_weights("zeromodels/seresnext101_32x8d_ah_in1k")

image = Image.open("your_image.jpg").convert("RGB")
pixels = processor(image)  # resize + normalize (normalization lives in the processor)
logits = model(pixels, training=False)
print(logits.shape)  # (1, num_classes)

# Feature extraction: the backbone without the classifier head
backbone = SENetModel.from_weights("zeromodels/seresnext101_32x8d_ah_in1k", as_backbone=True)
features = backbone(pixels, training=False)

Load any SENet variant the same way with from_weights("zeromodels/<variant>"):

Variant	Hub
`seresnet50_a1_in1k`	`zeromodels/seresnet50_a1_in1k`
`seresnext101_32x4d_gluon_in1k`	`zeromodels/seresnext101_32x4d_gluon_in1k`
`seresnext101_32x8d_ah_in1k`	`zeromodels/seresnext101_32x8d_ah_in1k`
`seresnext50_32x4d_gluon_in1k`	`zeromodels/seresnext50_32x4d_gluon_in1k`
`seresnext50_32x4d_racm_in1k`	`zeromodels/seresnext50_32x4d_racm_in1k`

Tips

Set KERAS_BACKEND before importing Keras / zeromodels.
SENetImageClassify returns class logits; SENetModel returns features (as_backbone=True for multi-scale stages).
See docs and Loading Weights.
Upstream / timm checkpoints: SENetImageClassify.from_weights("hf:timm/seresnext101_32x8d.ah_in1k").

Special Thanks

A huge thank you to the SENet authors and the timm / Hub communities for creating and releasing these models.

License: see YAML license (usually matches the upstream checkpoint).

kerasformers/seresnext101_32x8d_ah_in1k