v0.8.0

Update docs.

v0.7.6

Do not require `State` to extend `AsAny`.

v0.7.5

Bump version to v0.7.5

v0.7.4

Fix backing slot mismatch by also checking the job's info.

v0.7.3

Fix batch inference.

v0.7.2

Require `Model` to be `Send + Sync`.

v0.7.1

Bump version to v0.7.1

v0.7.0

Update README.md

v0.6.36

Allow automatically choose adapter in examples.

v0.6.35

Fix tensor init cache (#22)

* Implement init for `TensorGpu`.

* Re-implement resource cache.

* Simplify `TensorInitContext` trait.

* Bump version to v0.6.35