- github.com/alecthomas/template: [a0175ee → fb15b89](https://github.com/alecthomas/template/compare/a0175ee...fb15b89)
- github.com/alecthomas/units: [2efee85 → c3de453](https://github.com/alecthomas/units/compare/2efee85...c3de453)
- github.com/beorn7/perks: [v1.0.0 → v1.0.1](https://github.com/beorn7/perks/compare/v1.0.0...v1.0.1)

<sign>            ::= \"+\" | \"-\" <signedNumber>    ::= <number> | <sign><number> <suffix>          ::= <binarySI> | <decimalExponent> | <decimalSI> <binarySI>        ::= Ki | Mi | Gi | Ti | Pi | Ei\n\n\t(International System of units; See: http://physics.nist.gov/cuu/Units/binary.html)\n\n<decimalSI>       ::= m | \"\" | k | M | G | T | P | E\n\n\t(Note that 1024 = 1Ki but 1000 = 1k; I didn't choose the capitalization.)\n\n<decimalExponent> ::= \"e\" <signedNumber> | \"E\" <signedNumber> ```\n\nNo...

<sign>            ::= \"+\" | \"-\" <signedNumber>    ::= <number> | <sign><number> <suffix>          ::= <binarySI> | <decimalExponent> | <decimalSI> <binarySI>        ::= Ki | Mi | Gi | Ti | Pi | Ei\n\n\t(International System of units; See: http://physics.nist.gov/cuu/Units/binary.html)\n\n<decimalSI>       ::= m | \"\" | k | M | G | T | P | E\n\n\t(Note that 1024 = 1Ki but 1000 = 1k; I didn't choose the capitalization.)\n\n<decimalExponent> ::= \"e\" <signedNumber> | \"E\" <signedNumber> ```\n\nNo...

rfs($fs, $important: false) {\n  // Cache $fs unit\n  $fs-unit: if(type-of($fs) == \"number\", unit($fs), false);\n\n  // Add !important suffix if needed\n  $rfs-suffix: if($important, \" !important\", \"\");\n\n  // If $fs isn't a number (like inherit) or $fs has a unit (not px or rem, like 1.5em) or $ is 0, just print the value\n  @if not $fs-unit or $fs-unit != \"\" and $fs-unit != \"px\" and $fs-unit != \"rem\" or $fs == 0 {\n    font-size: #{$fs}#{$rfs-suffix};\n  }\n  @else {\n    // Variables for...

>>> tf.nn.elu(0.0)
<tf.Tensor: shape=(), dtype=float32, numpy=0.0>
>>> tf.nn.elu(-1000.0)
<tf.Tensor: shape=(), dtype=float32, numpy=-1.0>

See [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
](http://arxiv.org/abs/1511.07289)
  }];

  let arguments = (ins
    TF_FloatTensor:$features
  );

  let results = (outs
    TF_FloatTensor:$activations
  );

rfs($fs, $important: false) {\n  // Cache $fs unit\n  $fs-unit: if(type-of($fs) == \"number\", unit($fs), false);\n\n  // Add !important suffix if needed\n  $rfs-suffix: if($important, \" !important\", \"\");\n\n  // If $fs isn't a number (like inherit) or $fs has a unit (not px or rem, like 1.5em) or $ is 0, just print the value\n  @if not $fs-unit or $fs-unit != \"\" and $fs-unit != \"px\" and $fs-unit != \"rem\" or $fs == 0 {\n    font-size: #{$fs}#{$rfs-suffix};\n  }\n  @else {\n    // Variables for...

Bubble, by reason of some Inequalities in the subsiding of the Water.
And sometimes small black Spots generated at the sides would ascend up
to the larger black Spot at the top of the Bubble, and unite with it.

_Obs._ 18. Because the Colours of these Bubbles were more extended and
lively than those of the Air thinn'd between two Glasses, and so more

pkg syscall (freebsd-arm64), type RtMsghdr struct, Fmask int32
pkg syscall (freebsd-arm64), type RtMsghdr struct, Index uint16
pkg syscall (freebsd-arm64), type RtMsghdr struct, Inits uint64
pkg syscall (freebsd-arm64), type RtMsghdr struct, Msglen uint16
pkg syscall (freebsd-arm64), type RtMsghdr struct, Pad_cgo_0 [2]uint8
pkg syscall (freebsd-arm64), type RtMsghdr struct, Pid int32

10134..10136  ; disallowed                             # NA   <reserved-10134>..<reserved-10136>
10137..1013F  ; valid                  ;      ; NV8    # 4.0  AEGEAN WEIGHT BASE UNIT..AEGEAN MEASURE THIRD SUBUNIT
10140..1018A  ; valid                  ;      ; NV8    # 4.1  GREEK ACROPHONIC ATTIC ONE QUARTER..GREEK ZERO SIGN

        `tf.keras.optimizers.Optimizer`/`Adam`/etc will point to the new
        optimizers, and the previous generation of optimizers will be moved to
        `tf.keras.optimizers.legacy.Optimizer`/`Adam`/etc.
    *   Added L2 unit normalization layer `tf.keras.layers.UnitNormalization`.
    *   Added `tf.keras.regularizers.OrthogonalRegularizer`, a new regularizer
        that encourages orthogonality between the rows (or columns) or a weight