Swish activation function

README.md

@@ -1,6 +1,6 @@
 # PEtab SciML test suite
 
-[![Build Status](https://https://github.com/PEtab-dev/petab_sciml_testsuite/actions/workflows/CI.yml/badge.svg?branch=main)](https://https://github.com/PEtab-dev/petab_sciml_testsuite/actions/workflows/CI.yml?query=branch%3Amain)
+[![Build Status](https://github.com/PEtab-dev/petab_sciml_testsuite/actions/workflows/CI.yml/badge.svg?branch=main)](https://github.com/PEtab-dev/petab_sciml_testsuite/actions/workflows/CI.yml?query=branch%3Amain)
 [![Aqua QA](https://raw.githubusercontent.com/JuliaTesting/Aqua.jl/master/badge.svg)](https://github.com/JuliaTesting/Aqua.jl)
 [![code style: runic](https://img.shields.io/badge/code_style-%E1%9A%B1%E1%9A%A2%E1%9A%BE%E1%9B%81%E1%9A%B2-black)](https://github.com/fredrikekre/Runic.jl)
 
@@ -60,9 +60,9 @@ Currently, supported and tested layers and activation functions are:
 - **Layers**: `Conv1-3d`, `ConvTranspose1-3d`, `AvgPool1-3d`, `MaxPool1-3d`, `LPPool1-3d`,
   `AdaptiveMaxPool1-3d`, `AdaptiveMeanPool1-3d`, `BatchNorm1-3d`, `InstanceNorm1-3d`,
   `LayerNorm`, `Dropout1-3d`, `Dropout`, `AlphaDropout`, `Linear`, `Bilinear` and `Flatten`.
-- **Activation**: `relu`, `relu6`, `hardtanh`, `hardswish`, `selu`, `leaky_relu`, `gelu`,
-  `logsigmoid`, `tanhshrink`, `softsign`, `softplus`, `tanh`, `sigmoid`, `hardsigmoid`,
-  `mish`, `elu`, `celu`, `softmax` and `log_softmax`.
+- **Activation**: `relu`, `relu6`, `hardtanh`, `hardswish`, `selu`, `silu` `leaky_relu`,
+  `gelu`, `logsigmoid`, `tanhshrink`, `softsign`, `softplus`, `tanh`, `sigmoid`,
+  `hardsigmoid`, `mish`, `elu`, `celu`, `softmax` and `log_softmax`.
 
 A list of supported layers and activation functions for each tool supporting PEtab SciML can
 be found in the PEtab SciML

test_cases/ml_model_import/054/README.md

@@ -0,0 +1,7 @@
+# Net Test Case 054
+
+Test case for testing neural network import only.
+
+## Neural-net
+
+A simple feed-forward network testing `silu` (often called `swish`) activation function.

test_cases/ml_model_import/054/create_testdata/net.jl

@@ -0,0 +1,27 @@
+using Lux, StableRNGs
+using PEtabSciMLTestsuite: save_ps, save_io, write_yaml
+
+# runic: off
+nn_model = @compact(
+    layer1 = Dense(2, 5, Lux.swish),
+    layer2 = Dense(5, 1),
+) do x
+    embed = layer1(x)
+    out = layer2(embed)
+    @return out
+end
+# runic: on
+
+input_order_jl, input_order_py = ["W"], ["W"]
+output_order_jl, output_order_py = ["W"], ["W"]
+dirsave = joinpath(@__DIR__, "..")
+for i in 1:3
+    rng = StableRNG(i)
+    ps, st = Lux.setup(rng, nn_model)
+    input = rand(rng, Float32, 2)
+    output = nn_model(input, ps, st)[1]
+    save_ps(dirsave, i, nn_model, :net0, ps)
+    save_io(dirsave, i, input, input_order_jl, input_order_py, :input)
+    save_io(dirsave, i, output, output_order_jl, output_order_py, :output)
+end
+write_yaml(dirsave, input_order_jl, input_order_py, output_order_jl, output_order_py)

test_cases/ml_model_import/054/create_testdata/net.py

@@ -0,0 +1,26 @@
+"""Neural network import test generation"""
+
+import torch
+from torch import nn
+from torch.nn import functional as F
+from pysrc.ml_import_helper import make_yaml, test_nn
+
+
+class Net(nn.Module):
+    def __init__(self) -> None:
+        super().__init__()
+        self.layer1 = nn.Linear(2, 5)
+        self.layer2 = nn.Linear(5, 1)
+
+    def forward(self, net_input: torch.Tensor) -> torch.Tensor:
+        x = self.layer1(net_input)
+        x = F.silu(x)
+        out = self.layer2(x)
+        return out
+
+
+def ml_model_import_054(dir_save):
+    net = Net()
+    make_yaml(net, dir_save)
+    test_nn(net, dir_save, ["layer1", "layer2"])
+    return 0

test_cases/ml_model_import/054/net.yaml

@@ -0,0 +1,47 @@
+nn_model_id: net0
+inputs:
+- input_id: input0
+layers:
+- layer_id: layer1
+  layer_type: Linear
+  args:
+    in_features: 2
+    out_features: 5
+    bias: true
+- layer_id: layer2
+  layer_type: Linear
+  args:
+    in_features: 5
+    out_features: 1
+    bias: true
+forward:
+- name: net_input
+  op: placeholder
+  target: net_input
+  args: []
+  kwargs: {}
+- name: layer1
+  op: call_module
+  target: layer1
+  args:
+  - net_input
+  kwargs: {}
+- name: silu
+  op: call_function
+  target: silu
+  args:
+  - layer1
+  kwargs:
+    inplace: false
+- name: layer2
+  op: call_module
+  target: layer2
+  args:
+  - silu
+  kwargs: {}
+- name: output
+  op: output
+  target: output
+  args:
+  - layer2
+  kwargs: {}

test_cases/ml_model_import/054/solutions.yaml

@@ -0,0 +1,21 @@
+input_order_py:
+  - "W"
+net_input:
+  - "net_input_1.hdf5"
+  - "net_input_2.hdf5"
+  - "net_input_3.hdf5"
+input_order_jl:
+  - "W"
+output_order_jl:
+  - "W"
+net_ps:
+  - "net_ps_1.hdf5"
+  - "net_ps_2.hdf5"
+  - "net_ps_3.hdf5"
+net_file: "net.yaml"
+output_order_py:
+  - "W"
+net_output:
+  - "net_output_1.hdf5"
+  - "net_output_2.hdf5"
+  - "net_output_3.hdf5"