Alexey Gritsenko
Alexey Gritsenko
I checked a number of splits for a number CASPs - both in TFRecord and in textual formats. I wasn't exhaustive, but it seems like secondary structure data is missing...
Hi @moabarar Thank for figuring this out. Would you like to propose a PR for your fix? ~ Alexey
CLIP uses the byte pair tokeniser implemented here: https://github.com/openai/CLIP/blob/main/clip/simple_tokenizer.py The resulting tokens will not be characters.
Hi Alara, Many thanks for doing the work of implementing OWL-ViT in HuggingFace. This is really cool and an exciting thing for us. I joined the Google AI org on...
```python def get_mat(n, m = None): return np.random.normal(size=(n, m or n)) mat = get_mat(5000, 300) mat_padded = np.concatenate([mat, np.full((5000, 5000 - 300), 10.)], axis=1) %timeit scipy.optimize.linear_sum_assignment(mat) # 23.1 ms ±...
Attaching the [config gist](https://gist.github.com/AlexeyG/bad5ebe8520bd1fe45356125bb009bd4)