-
### System Info
- Platform: Linux-6.8.0-49-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- PyTorch version: 2.5.1
- CUDA device(s): NVIDIA GeForce RTX 4090
- Transformers version: 4.46.…
-
For the naive expression, we should pre-submit an alternatives call, then actually load the alternatives when the user clicks.
-
BerTopic Version = 0.16.4
```
umap_model = umap.UMAP(n_neighbors=15, n_components=24, min_dist=0.0, metric='cosine', random_state=100)
embedding_model = SentenceTransformer('sentence-transformers/p…
-
siibra-python (v1.0a19 an prior) fetches the affine neuroglrancer/precomputed images from `transform.json` files stored along with the images (`baseurl/transform.json`). However, these transform files…
-
There are two fundamental issues with our `post_explicit!` and `post_implicit!` hooks:
- First, the name is bad in that they convey when in the time marching algorithm they are called, and not wha…
-
Related to https://github.com/instructlab/sdg/issues/201
We decided not to move forward with the hugging face approach, but to have better results with SDG + train, we need to have a precomputed da…
-
There's a lot here so I'll do what I often do: This is an umbrella issue that lists a series of tasks needed for the precompute / background task feature. Each of the below items should be filed as a…
-
For #797, we created a `/precomputed/annotations` endpoint which gets or sets all line annotations within a bounding box. But we punted on the behavior of lines that touch but extend beyond the given…
-
### Describe the workflow you want to enable
I am trying to use precompute=True for LassoCV. To save memory, I am passing in the inputs as float32's. However, I get an error that the Gram matrix prec…
-
In order to improve skills training, we require a precomputed dataset that is going to be mixed with the new synthetically generated dataset. This is mostly used during full training via instructlab/t…