([#42873](https://github.com/kubernetes/kubernetes/pull/42873),  [@xilabao](https://github.com/xilabao))
* You can now display requests/limits of extended resources in node allocated resources. ([#46079](https://github.com/kubernetes/kubernetes/pull/46079), [@xiangpengzhao](https://github.com/xiangpengzhao))

- Kube-scheduler: in Kubernetes 1.33, the number of devices that can be allocated per ResourceClaim was accidentally reduced to 16. Now the supported number of devices per ResourceClaim is 32 again. ([#131679](https://github.com/kubernetes/kubernetes/pull/131679), [@mortent](https://github.com/mortent)) [SIG Node]

- Image garbage collection no longer fails for images with only one tag but more than one repository associated. ([#70647](https://github.com/kubernetes/kubernetes/pull/70647), [@corvus-ch](https://github.com/corvus-ch))
- Re-issued Allocate grpc calls before starting a container that requests device-plugin resources if the cached state is missing. ([#73824](https://github.com/kubernetes/kubernetes/pull/73824), [@jiayingz](https://github.com/jiayingz))

  Notes:
  1. Only the first two CIDRs are used (soft limits for Alpha, might be lifted later on).
  2. Only the "RangeAllocator" (default) is allowed as a value for `--cidr-allocator-type`. Cloud allocators are not compatible with IPv6 dual stack
     ([#73977](https://github.com/kubernetes/kubernetes/pull/73977), [@khenidak](https://github.com/khenidak))

- Add huge page stats to Allocated resources in "kubectl describe node" ([#80605](https://github.com/kubernetes/kubernetes/pull/80605), [@odinuge](https://github.com/odinuge)) [SIG CLI]

### Feature

- Add --dry-run to kubectl delete, taint, replace ([#88292](https://github.com/kubernetes/kubernetes/pull/88292), [@julianvmodesto](https://github.com/julianvmodesto)) [SIG CLI and Testing]
- Add huge page stats to Allocated resources in "kubectl describe node" ([#80605](https://github.com/kubernetes/kubernetes/pull/80605), [@odinuge](https://github.com/odinuge)) [SIG CLI]

* Backporting TPR fix to 1.4 ([#42380](https://github.com/kubernetes/kubernetes/pull/42380), [@foxish](https://github.com/foxish))
* Fix AWS device allocator to only use valid device names ([#41455](https://github.com/kubernetes/kubernetes/pull/41455), [@gnufied](https://github.com/gnufied))

- When using `kubectl replace` (or the equivalent API call) on a Service, the caller no longer needs to do a read-modify-write cycle to fetch the allocated values for `.spec.clusterIP` and `.spec.ports[].nodePort`.  Instead the API server will automatically carry these forward from the original object when the new object does not specify them. ([#104672](https://github.com/kubernetes/kubernetes/pull/104672),...

        `tf.function`-local variable: during each execution of the
        `tf.function`, the variable will be created and then disposed, similar
        to a local (i.e. stack-allocated) variable in C/C++. Currently
        `experimental_enable_variable_lifting=False` only works on non-XLA
        devices (e.g. under `@tf.function(jit_compile=False)`).

*   TF SavedModel:

- Removed the generally available feature gate `ProxyTerminatingEndpoints`. ([#122134](https://github.com/kubernetes/kubernetes/pull/122134), [@ty-dc](https://github.com/ty-dc))