v0.2.14

• Added support for LoRA training.

v0.2.13

• Adding VGG models.
• Fixing bugs in SAM model inference.

v0.2.12

• Fixing small bugs in SAM.

v0.2.11

• Added Segment Aything models.

v0.2.10

• Exposed pooled features for ConvNeXt.

v0.2.9

• Exposed attention map in ViT models.

v0.2.8

• tfimm now supports python 3.10.

v0.2.7

• Added EfficientNet and MobileNet-V2 models.

v0.2.6

• Added tiny and small ConvNeXt models (@MSchnei)

v0.2.5

• Preprocessing works for abritrary number of in_channels.
• Removed temporary version restriction on libclang.