One: [luci-interpreter] Operators to enable (2)

[jinevening]

Next step of #1157.

You can learn how to add operators to luci-interpreter from previous PRs.
(ex: Pad #1509, AveragePool2D #1218, Reshape #1099, FullyConnected #1087)

If you want to participate in operator implementation, pick an operator, put your name into the PR column and replace it with the link to the PR once you are done.

Top priority
Operators | Status (float) | Status (uint8) | PR
-- | :--: | :--: | --
logistic | done | done | #1810
transpose | done | done | #1858
transpose_conv | done | done | #2033

Long-term goal (in order from easy to difficult)

Activations | Status (float) | Status (uint8) | PR
-- | :--: | :--: | --
elu | done | — |#2147
leaky_relu | done | done |#2160
prelu | done | done | #4039
relu | done | done | #4150
relu6 | done | done | #4192
tanh | done | done | #3622

Element-wise | Status (float) | Status (uint8) | PR
-- | :--: | :--: | --
logical_or | only bool | only bool | #4716
div | done | done | #4425
equal | done | done | #4336
floor | done | — | #4402
floor_div(floordiv) | done | — | #4421
greater | done | done | #4336
greater_equal | done | done | #4336
less | done | done | #4336
less_equal | done | done | #4336
not_equal | done | done | #4336
pow | done | — | #4361
rsqrt | done | — | #3992
sqrt | done | — | #3904
sub(subtract) | done | done | #4464

Space transformation | Status (float) | Status (uint8) | PR
-- | :--: | :--: | --
depth_to_space | done | done | #3466
resize_bilinear | done | done | #4163
resize_nearestneighbor | done | done | #4106
reverse | done | done | #3435
slice | done | done | #3385
space_to_depth | done | done | #3230
split | done | done | #1742
squeeze | done | done | #3212
strided_slice | done | done | #3049

Other | Status (float) | Status (uint8) | PR
-- | :--: | :--: | --
l2_normalize | done | done |#2394
l2_pool | done | — |#2321
local_response_normalize | done | — |#2350

[struss]

I found that deconvolution is same with conv2d_transpose.
https://www.tensorflow.org/api_docs/python/tf/nn/conv2d_transpose

[jinevening]

@struss , I removed deconvolution in the table, because circle schema does not have deconv Op. Thanks.

[s-barannikov]

@jinevening
There is no relun op in TFLite and I can't find in in TensorFlow. Is it OK to remove it?

[jinevening]

There is no relun op in TFLite and I can't find in in TensorFlow. Is it OK to remove it?

Sure, thanks.

[s-barannikov]

@jinevening Everything seems to be done.

[jinevening]

Everything seems to be done.

Great! I didn't expect that we can finish the list so soon. Thanks for the hard works from SRR and @struss . 👍

It seems that logical_or was excluded from the list at some point. I'll close this issue after logical_or is implemented.