zmughal commented on a change in pull request #20852:
URL: https://github.com/apache/incubator-mxnet/pull/20852#discussion_r794252536
##########
File path: perl-package/AI-MXNet/lib/AI/MXNet/Metric.pm
##########
@@ -79,7 +79,7 @@ use overload '""' => sub {
has 'name' => (is => 'rw', isa => 'Str');
has 'num' => (is => 'rw', isa => 'Int');
has 'num_inst' => (is => 'rw', isa => 'Maybe[Int|ArrayRef[Int]]');
-has 'sum_metric' => (is => 'rw', isa => 'Maybe[Num|ArrayRef[Num]]');
+has 'sum_metric' => (is => 'rw', isa => 'Maybe[Num|ArrayRef[Num]|PDL]');
Review comment:
== examples/calculator.pl ==
```
fullyconnected4_bias -> [0.675975]
fullyconnected4_weight ->
[
[ 1.00116 0.999985]
]
fullyconnected5_bias -> [4.85674e-05]
fullyconnected5_weight ->
[
[0.000844207 0.000230625 0.508509]
]
fullyconnected6_bias -> [0.385759]
fullyconnected6_weight ->
[
[ 1 -1]
]
fullyconnected7_bias -> [-2.62715e-06]
fullyconnected7_weight ->
[
[-1.31198e-05 8.74287e-06 0.679938]
]
12345 + 54321 ≈ 66742.1796875
188 - 88 ≈ 99.8591766357422
250 * 2 ≈ 503.255798339844
250 / 2 ≈ 124.998970031738
```
== examples/mnist.pl ==
```
Epoch[4] Batch [200] Speed: 59678.68 samples/sec Train-accuracy=0.974726
Epoch[4] Batch [400] Speed: 57716.06 samples/sec Train-accuracy=0.973350
Epoch[4] Train-accuracy=0.974233
Epoch[4] Time cost=1.057
Epoch[4] Validation-accuracy=0.967800
Epoch[5] Batch [200] Speed: 52405.81 samples/sec Train-accuracy=0.977214
Epoch[5] Batch [400] Speed: 57804.15 samples/sec Train-accuracy=0.976100
Epoch[5] Train-accuracy=0.976950
Epoch[5] Time cost=1.079
Epoch[5] Validation-accuracy=0.966500
Epoch[6] Batch [200] Speed: 57043.78 samples/sec Train-accuracy=0.977015
Epoch[6] Batch [400] Speed: 57357.53 samples/sec Train-accuracy=0.975800
Epoch[6] Train-accuracy=0.976433
Epoch[6] Time cost=1.052
Epoch[6] Validation-accuracy=0.963000
Epoch[7] Batch [200] Speed: 45378.31 samples/sec Train-accuracy=0.977711
Epoch[7] Batch [400] Speed: 55753.34 samples/sec Train-accuracy=0.976350
Epoch[7] Train-accuracy=0.977467
Epoch[7] Time cost=1.144
Epoch[7] Validation-accuracy=0.963700
Epoch[8] Batch [200] Speed: 54546.75 samples/sec Train-accuracy=0.980448
Epoch[8] Batch [400] Speed: 55774.47 samples/sec Train-accuracy=0.980000
Epoch[8] Train-accuracy=0.980217
Epoch[8] Time cost=1.090
Epoch[8] Validation-accuracy=0.962900
Epoch[9] Batch [200] Speed: 48810.85 samples/sec Train-accuracy=0.980050
Epoch[9] Batch [400] Speed: 54945.51 samples/sec Train-accuracy=0.982550
Epoch[9] Train-accuracy=0.981850
Epoch[9] Time cost=1.115
Epoch[9] Validation-accuracy=0.969500
```
== examples/sparse/matrix_factorization/train.pl ==
```
Epoch[2] Batch [60300] Speed: 50291.13 samples/sec
Train-mse=0.706833
Epoch[2] Batch [60400] Speed: 54047.43 samples/sec
Train-mse=0.723078
Epoch[2] Batch [60500] Speed: 53646.70 samples/sec
Train-mse=0.741219
Epoch[2] Batch [60600] Speed: 52929.93 samples/sec
Train-mse=0.706861
Epoch[2] Batch [60700] Speed: 53583.14 samples/sec
Train-mse=0.734509
Epoch[2] Batch [60800] Speed: 50907.01 samples/sec
Train-mse=0.735880
Epoch[2] Batch [60900] Speed: 39899.38 samples/sec
Train-mse=0.758780
Epoch[2] Batch [61000] Speed: 48115.77 samples/sec
Train-mse=0.693772
Epoch[2] Batch [61100] Speed: 53833.52 samples/sec
Train-mse=0.714911
Epoch[2] Batch [61200] Speed: 53998.88 samples/sec
Train-mse=0.731971
Epoch[2] Batch [61300] Speed: 51153.75 samples/sec
Train-mse=0.732873
Epoch[2] Batch [61400] Speed: 51454.19 samples/sec
Train-mse=0.715944
Epoch[2] Batch [61500] Speed: 51353.83 samples/sec
Train-mse=0.715877
Epoch[2] Batch [61600] Speed: 53121.93 samples/sec
Train-mse=0.703896
Epoch[2] Batch [61700] Speed: 53345.31 samples/sec
Train-mse=0.700082
Epoch[2] Batch [61800] Speed: 48247.09 samples/sec
Train-mse=0.734545
Epoch[2] Batch [61900] Speed: 42918.29 samples/sec
Train-mse=0.714020
Epoch[2] Batch [62000] Speed: 49330.02 samples/sec
Train-mse=0.722819
Epoch[2] Batch [62100] Speed: 53999.32 samples/sec
Train-mse=0.736258
Epoch[2] Batch [62200] Speed: 51953.96 samples/sec
Train-mse=0.737786
Epoch[2] Batch [62300] Speed: 51416.39 samples/sec
Train-mse=0.698692
Epoch[2] Batch [62400] Speed: 53206.32 samples/sec
Train-mse=0.716579
Epoch[2] Batch [62500] Speed: 52938.28 samples/sec
Train-mse=0.733243
Preparing data iterators for ./data/ml-10M100K/r1.train ...
Preparing data iterators for ./data/ml-10M100K/r1.test ...
Training started ...
epoch 0, eval MSE = 0.99958598613739
epoch 1, eval MSE = 1.03603255748749
epoch 2, eval MSE = 1.07017529010773
Training completed.
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
