Custom Activation dataset #46
Description
Hi when I try to load my custom activation dataset I keep running this error
/usr/local/lib/python3.11/dist-packages/dictionary_learning/trainers/top_k.py in encode(self, x, return_topk, use_threshold)
82
83 def encode(self, x: t.Tensor, return_topk: bool = False, use_threshold: bool = False):
---> 84 post_relu_feat_acts_BF = nn.functional.relu(self.encoder(x - self.b_dec))
85
86 if use_threshold:
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!
which indicates some of the model is not moved to the GPU. Has this code been tested ?