-
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
Traceback (most recent call last):
File "/Vary/run_opt.py", line 111, in
eval_m…
-
### What version of Bun is running?
1.1.22
### What platform is your computer?
Microsoft Windows NT 10.0.22631.0 x64 (also linux on fedora 40)
### What steps can reproduce the bug?
Run …
Miosp updated
1 month ago
-
Not sure if this has been explored before.
The reason why monad transformers couldn't be covariant in their result / extra types was mostly that if the transformer were to be covariant, we would ne…
-
## Overview
In terms of the model of core logic in Plangoab, we implemented using not a class but plain JS object because of reason as part of some experiment.
However, I finally realized this wa…
-
We have anecdotal evidence that the Python layer in between libtorch and Postgres calls can increase the cost as much as 4x when doing large batches. In the end state, I think we should be boiling the…
-
For any `MonadLogic` instance, we can write
```haskell
gather :: MonadLogic m => m a -> m [a]
gather m = msplit m >>= \case
Nothing -> pure []
Just (a, m') -> (a:) gather m'
```
Howeve…
-
#### Problem
Currently the `HasForward` typeclass splits the implementation of the forward pass into 2 functions:
- `forward :: f -> a -> b`, for a forward pass that is a pure function, and
- `f…
-
## 🚀 Feature
I propose adding instead of batch size a dictionary with batch size per GPU, for example {"cuda0":4, "cuda1", 6}
### Motivation
I have gv100 (32gb) and 3090 (24gb). using the cur…
-
Ohad asked me at ICFP to try to come up with a tiny example and post this here. The larger context is described at https://unsafeperform.io/blog/2022-07-02-a_small_benchmark_for_functional_languages_t…
-
When creating a highly nested, but concrete type and using it within a `do` statement, idris2 generates very long `ss` files, and typechecking and compilation becomes very slow.
Not sure if this is…