Daniel Gafni
Daniel Gafni
### Dagster version 1.2.2 ### What's the issue? Why does `SourceAsset` take `io_manager_def` or `io_manager_key` as input, but doesn't support `input_manager_def` or `input_manager_key`? I'm writing `InputManager`s specifically for my `SourceAsset`s,...
# Environment **Delta-rs version**: 0.17.3 **Binding**: Python **Environment**: Kubernetes (EKS) - **Cloud provider**: AWS - **OS**: Ubuntu - **Other**: *** # Bug **What happened**: It seems like `deltalake` fails to...
[Docos] Add new FAQ entry: "Type-check framework-agnostic tensors with `jaxtyping`! It is glorious!"
Passing a normal `numpy` array to `@jaxtyped(typechecker=beartype)`-decorated function raises an error. Passing a `jax.numpy` array of the same shape and type doesn't. ```python from beartype import beartype import numpy as...
### Search before asking - [X] I had searched in the [issues](https://github.com/ray-project/kuberay/issues) and found no similar feature requirement. ### Description Currently it's only possible to install it from git, which...
## Summary & Motivation This log message was accidentally removed in my last dagster-polars PR. Fixing this. ## How I Tested These Changes
At the time of writing the Datasets I didn't know about Samplers. They should be used now.
improve "fixed" visualization dataloaders on validation and test - currently they only output one image with different hue shifts. Should output different images
Adaptive instance normalization should remove some of the generator artifacts (like random color stains)