Daniel Vainsencher

### Description Before the fix, works only when `interpretation["attention"]` matches `encoder_length` (in which case the slice is not needed anyway). After, length matches always. I'm running into this with encoder_length=0,...

Update TotalVariation.jl to run, defining benchmark/Project.toml.

1

Might become a PR to bring more benchmarks to running state, feedback welcome.

Currently, just removes default padding. Plan for implementing padding versions is by extending the iterator with zeros, then calling the same underlying method. I propose for the default to require...

We shouldn't implicitly insert default values?

4

Matrix::from_iter\* etc all fill out the rest of the matrix with zeros if they run out of values early. I have several problems with this: - It masks bugs; I...

Things I've had to look up starting to use SciRust

9

I've contributed a little code, so I'm not as fresh as a typical new user, but I've still had to hunt around for the follow as I tried to convert...

Everything is unsafe; use slice instead?

11

Hi, at its core, the implementation of matrix is using a ptr, which then requires unsafe even to do a read only access of data (see, for example, the implementation...

nbdev_test fails when notebook imports polars package due to `set_start_method` conflict

1

# Failure and repro Trivial notebook/nbdev repo with `import polars as pl` fails ([repro CI run](https://github.com/daniel-vainsencher/nbdev-polars-mp-bug-repro/actions/runs/3329229804/jobs/5506182947)) with error message: ``` ImportError: Polars only works with python multiprocessing method set to:...

Add a projection to the probability simplex, and a test for it.

4

Useful for implementing gradient projection type algorithms on the probability simplex.

spans inside frames inside spans

10

I am not writing a game, but numerical code, such that algorithms have loops with some interesting parts inside the loop and some outside. So I want a span before...

Current support for LoRa is to compute the adjusted weights at loading time. Changing the set of LoRa layers applied requires reloading the original weights (W) Since LoRa components (A,...