-
Sometimes I have millions of data points to plot, e.g., with `scatter`. This is extremely inefficient so I first sparsity them by throwing away points that are close within some tolerance of already e…
-
### Problem
I'm always frustrated when I am trying to display the multiindex dataframe, but it is not readable
### Solution
`pandas` when displaying a dataframe that have multiindex skips dup…
-
I had a discussion with @soumith about this on Friday, but I wanted to record things down here so that I don't forget.
**Background.** Imagine that you are @martinraison and you are implementing sp…
-
For the sake of symmetry between SX/MX, we need to have sparsify not throw an error, but be an identity transformation for MX
-
Needed for keeping solver fast (high batch size), but prevent OOM in `sparsify` (need low batch size), see https://github.com/theislab/cellrank/issues/1146#issuecomment-1856213099
MUCDK updated
7 months ago
-
When using the recommended CentOS 9 image, it doesn't appear that `virt-sparsify` is available, which I've found leads to an increase in the amount of time needed for `os-migrate` to perform the workl…
-
Hi, I'm trying to sparsify my model using your code, but I'm encountering the "ExpectedMoreSplits Error" shown below.
The error arises from this line when loading the training data.
Could you he…
-
Hi,
I was wondering if you plan to put in a public domain the sparsified Llama2 models. In particular I am interested in the Llama2-70B with 50% unstructured sparsity.
Thanks!
egeor updated
11 months ago
-
- [x] Prove that `sparsify(tensor, threshold)` performs as intended
- [x] Prove `shrink` on a given layer produces the correct count of neurons for a given layer
- [x] ... for __all__ layers of a…
-
### Description
If the sparse matrix is passed as a parameter, I get an error when calling my normalize function. The function is fine otherwise
```python
import jax
import jax.numpy as jnp
d = j…