utils.caching

utils.caching

Functions

Name	Description
cache	Caching decorator - a local authoring convenience, use with care!
clear_all_cache	Clear all caches in the current cache root.
clear_experiment_cache	Clear the cache for the current experiment / notebook.
set_cache_path	Set the path where the cache is stored, relative to the cache root which lies next to this file in /cache.

cache

utils.caching.cache(fname='', redo=False)

Caching decorator - a local authoring convenience, use with care!

On the first call the decorated function runs and its return value is pickled to {fname}.pkl under cache_path; later calls load the pickle instead of recomputing.

The cache is meant to speed up local iteration, not to be a durable artifact: the pickles store live JAX/equinox objects and can go stale across library upgrades. To stay robust the loader recomputes (instead of crashing) whenever the pickle fails to load or its environment stamp (jax/tvboptim version, function source) no longer matches. For committed docs, prefer Quarto’s freeze over relying on these pickles.

fname: name of the file where the computation is stored under cache_path
redo: if True, the computation is run even if the file already exists

clear_all_cache

utils.caching.clear_all_cache()

Clear all caches in the current cache root.

clear_experiment_cache

utils.caching.clear_experiment_cache()

Clear the cache for the current experiment / notebook.

set_cache_path

utils.caching.set_cache_path(experiment='')

Set the path where the cache is stored, relative to the cache root which lies next to this file in /cache.

experiment: name of the experiment / notebook -> becomes a subdirectory in the cache root

# utils.caching { #tvboptim.utils.caching }

`utils.caching`



## Functions

| Name | Description |
| --- | --- |
| [cache](#tvboptim.utils.caching.cache) | Caching decorator - a local authoring convenience, use with care! |
| [clear_all_cache](#tvboptim.utils.caching.clear_all_cache) | Clear all caches in the current cache root. |
| [clear_experiment_cache](#tvboptim.utils.caching.clear_experiment_cache) | Clear the cache for the current experiment / notebook. |
| [set_cache_path](#tvboptim.utils.caching.set_cache_path) | Set the path where the cache is stored, relative to the cache root which lies next to this file in /cache. |

### cache { #tvboptim.utils.caching.cache }

```python
utils.caching.cache(fname='', redo=False)
```

Caching decorator - a local authoring convenience, use with care!

On the first call the decorated function runs and its return value is pickled to
`{fname}.pkl` under `cache_path`; later calls load the pickle instead of recomputing.

The cache is meant to speed up local iteration, not to be a durable artifact: the
pickles store live JAX/equinox objects and can go stale across library upgrades. To
stay robust the loader recomputes (instead of crashing) whenever the pickle fails to
load or its environment stamp (jax/tvboptim version, function source) no longer
matches. For committed docs, prefer Quarto's freeze over relying on these pickles.

* `fname`: name of the file where the computation is stored under `cache_path`
* `redo`: if True, the computation is run even if the file already exists

### clear_all_cache { #tvboptim.utils.caching.clear_all_cache }

```python
utils.caching.clear_all_cache()
```

Clear all caches in the current cache root.

### clear_experiment_cache { #tvboptim.utils.caching.clear_experiment_cache }

```python
utils.caching.clear_experiment_cache()
```

Clear the cache for the current experiment / notebook.

### set_cache_path { #tvboptim.utils.caching.set_cache_path }

```python
utils.caching.set_cache_path(experiment='')
```

Set the path where the cache is stored, relative to the cache root which lies next to this file in /cache.

* `experiment`: name of the experiment / notebook -> becomes a subdirectory in the cache root