* Replace `LossScaleOptimizer(opt, "dynamic")` with `LossScaleOptimizer(opt)`. If you pass anything other than `"dynamic"` to the second argument, see (1) of the next section.
    * In the following rare cases, you need to make more changes when switching to the non-experimental API:
      * If you passed anything other than `"dynamic"` to the `loss_scale` argument (the second argument) of `LossScaleOptimizer`:

approved service transitions to indicatorTrue. Invoking a non-approved // service transitions to indicatorFalse, and it can't leave that state until a // reset. The idea is that functions can "delegate" checks to inner functions, // and if there's anything non-approved in the stack, the final result is // negative. Finally, we expose indicatorUnset as negative to the user, so that // we don't need to explicitly annotate fully non-approved services. //go:linkname getIndicator crypto/internal/fips140.getIndicator...

approved service transitions to indicatorTrue. Invoking a non-approved // service transitions to indicatorFalse, and it can't leave that state until a // reset. The idea is that functions can "delegate" checks to inner functions, // and if there's anything non-approved in the stack, the final result is // negative. Finally, we expose indicatorUnset as negative to the user, so that // we don't need to explicitly annotate fully non-approved services. //go:linkname getIndicator crypto/internal/fips140.getIndicator...

* ⬆ Upgrade MkDocs Material and configs. PR [#4385](https://github.com/tiangolo/fastapi/pull/4385) by [@tiangolo](https://github.com/tiangolo).

## 0.70.1

There's nothing interesting in this particular FastAPI release. It is mainly to enable/unblock the release of the next version of Pydantic that comes packed with features and improvements. 🤩

### Fixes