*   Unary GPU kernels: Abs, Atanh, Acos, Acosh, Asin, Asinh, Atan, Cos, Cosh, Sin, Sinh, Tan, Tanh.
    *   Binary GPU kernels: AddV2, Sub, Div, DivNoNan, Mul, MulNoNan, FloorDiv, Equal, NotEqual, Greater, GreaterEqual, LessEqual, Less.

* `tf.lite`
    * Add experimental supports conversion of models that may be larger than 2GB before buffer deduplication

### Bug Fixes and Other Changes

- UDP and SCTP protocols can left stale connections that need to be cleared to avoid services disruption, but they can cause problems that are hard to debug.
  Kubernetes components using a loglevel greater or equal than 4 will log the conntrack operations and its output, to show the entries that were deleted. ([#95694](https://github.com/kubernetes/kubernetes/pull/95694), [@aojea](https://github.com/aojea)) [SIG Network]

- The `kubectl debug` is able to create ephemeral containers in pre-1.22 clusters with the `EphemeralContainers` feature enabled. Note that versions of kubectl prior to 1.22 are unable to create ephemeral containers in clusters version 1.22 and greater due to an API change. ([#103292](https://github.com/kubernetes/kubernetes/pull/103292), [@verb](https://github.com/verb))

this._addAriaAndCollapsedClass([element], this._isShown(selected))\n      }\n    }\n  }\n\n  _getFirstLevelChildren(selector) {\n    const children = SelectorEngine.find(CLASS_NAME_DEEPER_CHILDREN, this._config.parent)\n    // remove children if greater depth\n    return SelectorEngine.find(selector, this._config.parent).filter(element => !children.includes(element))\n  }\n\n  _addAriaAndCollapsedClass(triggerArray, isOpen) {\n    if (!triggerArray.length) {\n      return\n    }\n\n    for (const...

- Fixed a regression since 1.27.0 in scheduler framework when running score plugins. 
  The `skippedScorePlugins` number might be greater than `enabledScorePlugins`, 
  so when initializing a slice the cap(len(skippedScorePlugins) - len(enabledScorePlugins)) is negative, 

FE63          ; mapped                 ; 002D          # 1.1  SMALL HYPHEN-MINUS
FE64          ; disallowed_STD3_mapped ; 003C          # 1.1  SMALL LESS-THAN SIGN
FE65          ; disallowed_STD3_mapped ; 003E          # 1.1  SMALL GREATER-THAN SIGN
FE66          ; disallowed_STD3_mapped ; 003D          # 1.1  SMALL EQUALS SIGN
FE67          ; disallowed                             # NA   <reserved-FE67>

### **Storage**

*   [stable] `PersistentVolume` and `PersistentVolumeClaim` objects must now have a capacity greater than zero.
*   [stable] Mutation of `PersistentVolumeSource` after creation is no longer allowed
*   [alpha] Deletion of `PersistentVolumeClaim` objects that are in use by a pod no longer permitted (if alpha feature is enabled).

- Readjust `kubelet_containers_per_pod_count` buckets to only show metrics greater than 1. ([#98169](https://github.com/kubernetes/kubernetes/pull/98169), [@wawa0210](https://github.com/wawa0210))

on screen height and width\n$rfs-two-dimensional: false !default;\n\n// Factor of decrease\n$rfs-factor: 10 !default;\n\n@if type-of($rfs-factor) != number or $rfs-factor <= 1 {\n  @error \"`#{$rfs-factor}` is not a valid  $rfs-factor, it must be greater than 1.\";\n}\n\n// Mode. Possibilities: \"min-media-query\", \"max-media-query\"\n$rfs-mode: min-media-query !default;\n\n// Generate enable or disable classes. Possibilities: false, \"enable\" or \"disable\"\n$rfs-class: false !default;\n\n// 1...

- Update kube-apiserver's priority & fairness work estimator such that 'max seats' is MIN(0.15 x nominalCL, nominalCL / handSize)
  
  This fixes a bug where clients with requests using hand size x max seats greater than the nominal concurrency limit can starve other requests in the same priority level. ([#118601](https://github.com/kubernetes/kubernetes/pull/118601), [@andrewsykim](https://github.com/andrewsykim)) [SIG API Machinery]