Daniël de Kok c841a6c90d Improve mode handling (#108) ...

* Set `kernelize` default mode to `Mode.TRAINING | Mode.TORCH_COMPILE`

Also update docs and tests.

* Rename `Mode.DEFAULT` to `Mode.FALLBACK`

* More fine-grained fallbacks

For instance, INFERENCE can fall back to INFERENCE | TORCH_COMPILE,
TRAINING, TRAINING | TORCH_COMPILE, and FALLBACK.

* Update documtenation for mode fallback

* Mention that you can rerun `kernelize` to change the mode

2025-07-15 16:10:43 +02:00

..

kernel_locking

Make the forward pass torch.compile compatible (#87)

2025-06-03 15:06:02 +02:00

conftest.py

Add support for Metal builds (#89)

2025-05-30 15:54:28 +02:00

test_basic.py

Add get_local_kernel function (#102)

2025-07-01 13:58:47 +02:00

test_benchmarks.py

Add support for Metal builds (#89)

2025-05-30 15:54:28 +02:00

test_interval_tree.py

Support registering layers with a range of CUDA capabilities (#106)

2025-07-14 16:59:21 +02:00

test_kernel_locking.py

Add support for Metal builds (#89)

2025-05-30 15:54:28 +02:00

test_layer.py

Improve mode handling (#108)

2025-07-15 16:10:43 +02:00