-
https://github.com/neelnanda-io/TransformerLens/blob/ce82675a8e89b6d5e6229a89620c843c794f3b04/transformer_lens/loading_from_pretrained.py#L1395C7-L1395C69
runpod/pytorch:2.1.0-py3.10-cuda11.8.0-dev…
-
Traceback (most recent call last):
File "/workspace/./runner/train.py", line 569, in
main()
File "/workspace/./runner/train.py", line 564, in main
trainer = AF3Trainer(configs)
Fil…
-
Hi Romeric,
thanks for your great work!!! I am using your library for my project. However, I met some problems when using Einstain summation for tensor contraction. I wanted to do the similar calcula…
-
Some autogenerated resolvers look too complex, i.e. `resolve_operator__onnx__maxpool__12`. We don't need that many cases, just 5 "tensor(float16), tensor(float), tensor(double), tensor(int8), tensor(u…
-
### 🐛 Describe the bug
Hi,
When creating a tensor from a non-contiguous NumPy array, e.g.
```python
my_array = np.reshape(np.arange(10), (2, 5))
my_array = np.transpose(my_array, (1, 0))
my_te…
-
I am trying to train similar model for action classification on videos, It has around 200 classes.
Following is the code for loss_labels:
```py
def loss_labels(self, outputs, targets, indices, nu…
-
I had some problems using `transforms.Resize((32, 32)` because my network is designed for 3*32*32 images, so I tried to use `transforms.Resize((32, 32)`. But an error like this occurred:
```
TypeErr…
-
(This issue might be updated as our discussion evolves)
# Context
Broadcast IterDomains signify that a future operation might perform broadcasting involving this dimension. However, broadcasting…
-
I try to test my codes which calculate sdr with your separate samples(ex_18).
In my sdr codes, the result is about 6.47 while yours is 19.37.
can you help me find out anything wrong in my code…
-
Hi, @tridao . Yes, it's me again.
First, thanks for the [monarch paper](https://arxiv.org/pdf/2204.00595.pdf), it was quite a read ;)
I've been looking through the repo, but could not find the off…