Open LucFrachon opened 6 months ago
@LucFrachon Same idea, I have. Do you have any updates on your side ? Thanks !
I wrote code to update the state dict and it worked. I haven't checked if the provided checkpoints have been updated. I've moved on to other things now...
The weights for ConvNeXt-V2-Base, pretrained on INet1k, provided here, have several incompatibilities with the encoder architecture:
self.ln
element that means that the checkpoint keys should look likedownsample_layers.1.0.ln.bias
, whereas the weights provided havedownsample_layers.1.0.bias
.pwconv<i>
haveself.linear
, which the checkpoints doesn't have.kernel
(e.g.,downsample_layers.1.1.kernel
), whereas the provided checkpoint has keys"downsample_layers.1.1.weight
.stages
: biases need to be (1, c) but are (c,), and weights need to be (p ** 2, c) but are (c, -1, p, p)These can all be fixed with some code, but it would make life easier for everyone if you could upload the correct weights. Thanks a lot!