Metadata-Version: 2.4
Name: mura
Version: 0.2.19
Summary: WandB, Git, and run-table integration for ML++ / eliminating workflow unevenness
Author-email: Akhil Sadam <sadam.akhil@gmail.com>
Requires-Python: >=3.10
Description-Content-Type: text/markdown
Requires-Dist: wandb>=0.20.1
Requires-Dist: gitpython>=3.1.44
Requires-Dist: numpy>=2.0.0
Requires-Dist: lightning>=2.5.1
Requires-Dist: torch>=2.4.1
Requires-Dist: einops>=0.8.0
Requires-Dist: tqdm>=4.66.5
Requires-Dist: matplotlib>=3.7.1
Requires-Dist: pyyaml>=6.0.2
Requires-Dist: pydantic>=2.11.5
Requires-Dist: toml>=0.10.2
Requires-Dist: omegaconf>=2.3.0
Requires-Dist: pynvml>=12.0.0

## Changelog:
- See next-to-top version for stable release (latest is always WIP/unstable)

### 0.2.19 (2025-08-01)
- Add copy function for models

### 0.2.17 (2025-07-17)
- Add load from file for models

### 0.2.15 (2025-07-16)
- Changed GPUs to only include "free" GPUs; temporarily available due to cluster.
- Added missing pynvml dependency
- Added memory limit for GPU selection

### 0.2.12 (2025-07-16)
- Added full config info to model initialization (lightning run script)
- Added model adjust option (lightning run script)

### 0.2.9 (2025-07-15)
- Minor changes to lightning run script that matches the delay-neural-operator code


### 0.2.8 (2025-07-14)
- Major (breaking) change: Removed OmegaConf dependency except for `MISSING` value
- Added toml save support for configuration files
- TODO: add toml load support

### 0.2.6 (2025-07-13)
- Major (breaking) change: Refactor-in-progress to use `dataclasses` for better performance and readability
- Major (breaking) change: Ignoring cluster-job-deployment; this will be reintroduced in a future release
- TODO: add support for schema validation

### 0.1.0 (2025-01-01)
- Initial release with slow, dictionary-to-class parsing
- Note-to-self: older qg package releases need to be pinned to this version for compatibility
