-
### Describe the workflow you want to enable
Currently, there is no included implementation of a PCA algorithm made for handling binary data in the scikit-learn library. However, the algorithm for …
-
Continuing from #1461...
We attempt to avoid zero or negatives in the `log` (or division be zero or negatives in the gradient) by thresholding the quotient. Unfortunately, the code is not consisten…
-
请问,目前我运行了你的相关代码,我心中存在以下的疑惑:
(1)步骤二:
```
CKPT=105
TRAINING_DATA_NAME=dolly
TRAINING_DATA_FILE=../data/train/processed/dolly/dolly_data.jsonl # when changing data name, change the data path accordi…
-
**Describe the bug**
Equirectangular Projection have dark poles
**To Reproduce**
Steps to reproduce the behavior:
1. Set projection to Equirectangular Projection
2. Whait for render
3. See da…
-
First of all, thank you for developing torchtune. This has been very helpful for our group with limited GPU credits. I'm impressed by its capabilities, particularly its memory efficiency. I've noticed…
-
Here we will outline research project ideas for the postdocs
-
Inplace versions for projection_on_set, projection_gradient_on_set
-
### 🐛 Describe the bug
I believe `cholesky_backward` implementation is incorrect due to the false but well-intentioned assumption that $\overline{\Sigma}=\frac{S + S^\top}{2}$ is the backward gradi…
-
Hi, is the gradient scaling of `MLP` correct w.r.t. `expansion_factor`?
I noticed that in [demo.ipynb](https://github.com/graphcore-research/unit-scaling/blob/50ea961f1c0f2939a4bd7c8b9ada41acf9a583…
-
Hi Anthony,
Do you have the derivation steps of this jacobian matrix or any material/links for better understanding?
Thanks a lot !
https://github.com/hyperlogic/splatapult/blob/d05e3699384f82c7f1…