Renamed 'has_batch' to 'model_expects_batch_dim'.

README.md

@@ -38,7 +38,7 @@ On MacOS with M1 chip you will have to compile [tera_renderer](https://github.co
 and place the binary in `l4casadi/template_generation/bin`. For other platforms it will be downloaded automatically.
 
 ## Example
-https://github.com/Tim-Salzmann/l4casadi/blob/d800ca7bea785a22b05c64f4aecd7554ba5093d3/examples/readme.py#L28-L40
+https://github.com/Tim-Salzmann/l4casadi/blob/23e07380e214f70b8932578317aa373d2216b57e/examples/readme.py#L28-L40
 
 Please note that only `casadi.MX` symbolic variables are supported as input.
 

examples/acados.py

@@ -174,7 +174,7 @@ def acados_model(self, model):
 
 def run():
     N = 10
-    learned_dyn_model = l4c.L4CasADi(MultiLayerPerceptron(), has_batch=True, name='learned_dyn')
+    learned_dyn_model = l4c.L4CasADi(MultiLayerPerceptron(), model_expects_batch_dim=True, name='learned_dyn')
 
     model = DoubleIntegratorWithLearnedDynamics(learned_dyn_model)
     solver = MPC(model=model.model(), N=N,

examples/readme.py

@@ -25,7 +25,7 @@ def forward(self, x):
 
 
 pyTorch_model = MultiLayerPerceptron()
-l4c_model = l4c.L4CasADi(pyTorch_model, has_batch=True, device='cpu')  # device='cuda' for GPU
+l4c_model = l4c.L4CasADi(pyTorch_model, model_expects_batch_dim=True, device='cpu')  # device='cuda' for GPU
 
 x_sym = cs.MX.sym('x', 2, 1)
 y_sym = l4c_model(x_sym)

l4casadi/l4casadi.py

@@ -19,15 +19,25 @@ def dynamic_lib_file_ending():
 
 
 class L4CasADi(object):
-    def __init__(self, model: Callable[[torch.Tensor], torch.Tensor],
-                 has_batch: bool = False, device: Union[torch.device, Text] = "cpu", name: Text = "l4casadi_f"):
+    def __init__(self,
+                 model: Callable[[torch.Tensor], torch.Tensor],
+                 model_expects_batch_dim: bool = False,
+                 device: Union[torch.device, Text] = "cpu",
+                 name: Text = "l4casadi_f"):
+        """
+        :param model: PyTorch model.
+        :param model_expects_batch_dim: True if the PyTorch model expects a batch dimension.
+        :param device: Device on which the PyTorch model is executed.
+        :param name: Unique name of the generated L4CasADi model. This name is used for autogenerated files.
+            Creating two L4CasADi models with the same name will result in overwriting the files of the first model.
+        """
         self.model = model
         if isinstance(self.model, torch.nn.Module):
             self.model.eval().to(device)
             for parameters in self.model.parameters():
                 parameters.requires_grad = False
         self.name = name
-        self.has_batch = has_batch
+        self.has_batch = model_expects_batch_dim
         self.device = device if isinstance(device, str) else f'{device.type}:{device.index}'
 
         self.generation_path = pathlib.Path('./_l4c_generated')
@@ -100,7 +110,7 @@ def generate_cpp_function_template(self, rows: int, cols: int, has_jac: bool, ha
             'cols_out': cols_out,
             'has_jac': has_jac,
             'has_hess': has_hess,
-            'has_batch': self.has_batch
+            'model_expects_batch_dim': self.has_batch
         }
         with open(self.generation_path / f'{self.name}.json', 'w') as f:
             json.dump(gen_params, f)

l4casadi/template_generation/c_templates_tera/casadi_function.in.cpp

@@ -1,6 +1,6 @@
 #include <l4casadi.hpp>
 
-L4CasADi l4casadi("{{ model_path }}", "{{ name }}", {{ has_batch }}, "{{ device }}", {{ has_jac }}, {{ has_hess }});
+L4CasADi l4casadi("{{ model_path }}", "{{ name }}", {{ model_expects_batch_dim }}, "{{ device }}", {{ has_jac }}, {{ has_hess }});
 
 #ifdef __cplusplus
 extern "C" {

libl4casadi/include/l4casadi.hpp

@@ -8,9 +8,9 @@
 class L4CasADi
 {
 private:
-    bool has_batch;
+    bool model_expects_batch_dim;
 public:
-    L4CasADi(std::string, std::string, bool has_batch = false, std::string = "cpu", bool = false, bool = false);
+    L4CasADi(std::string, std::string, bool = false, std::string = "cpu", bool = false, bool = false);
     ~L4CasADi();
     void forward(const double*, int, int, double*);
     void jac(const double*, int, int, double*);

libl4casadi/src/l4casadi.cpp

@@ -79,14 +79,14 @@ class L4CasADi::L4CasADiImpl
     }
 };
 
-L4CasADi::L4CasADi(std::string model_path, std::string model_prefix, bool has_batch, std::string device,
+L4CasADi::L4CasADi(std::string model_path, std::string model_prefix, bool model_expects_batch_dim, std::string device,
         bool has_jac, bool has_hess):
     pImpl{std::make_unique<L4CasADiImpl>(model_path, model_prefix, device, has_jac, has_hess)},
-    has_batch{has_batch} {}
+    model_expects_batch_dim{model_expects_batch_dim} {}
 
 void L4CasADi::forward(const double* in, int rows, int cols, double* out) {
     torch::Tensor in_tensor;
-    if (this->has_batch) {
+    if (this->model_expects_batch_dim) {
         in_tensor = torch::from_blob(( void * )in, {1, rows}, at::kDouble).to(torch::kFloat);
     } else {
         in_tensor = torch::from_blob(( void * )in, {rows, cols}, at::kDouble).to(torch::kFloat);
@@ -98,7 +98,7 @@ void L4CasADi::forward(const double* in, int rows, int cols, double* out) {
 
 void L4CasADi::jac(const double* in, int rows, int cols, double* out) {
     torch::Tensor in_tensor;
-    if (this->has_batch) {
+    if (this->model_expects_batch_dim) {
         in_tensor = torch::from_blob(( void * )in, {1, rows}, at::kDouble).to(torch::kFloat);
     } else {
         in_tensor = torch::from_blob(( void * )in, {rows, cols}, at::kDouble).to(torch::kFloat);
@@ -110,7 +110,7 @@ void L4CasADi::jac(const double* in, int rows, int cols, double* out) {
 
 void L4CasADi::hess(const double* in, int rows, int cols, double* out) {
     torch::Tensor in_tensor;
-    if (this->has_batch) {
+    if (this->model_expects_batch_dim) {
         in_tensor = torch::from_blob(( void * )in, {1, rows}, at::kDouble).to(torch::kFloat);
     } else {
         in_tensor = torch::from_blob(( void * )in, {rows, cols}, at::kDouble).to(torch::kFloat);

tests/test_l4casadi.py

@@ -50,7 +50,7 @@ def test_l4casadi_deep_model(self, deep_model):
         rand_inp = torch.rand((1, deep_model.input_layer.in_features))
         torch_out = deep_model(rand_inp)
 
-        l4c_out = l4c.L4CasADi(deep_model, has_batch=True)(rand_inp.transpose(-2, -1).detach().numpy())
+        l4c_out = l4c.L4CasADi(deep_model, model_expects_batch_dim=True)(rand_inp.transpose(-2, -1).detach().numpy())
 
         np.allclose(l4c_out, torch_out.transpose(-2, -1).detach().numpy())
 
@@ -68,7 +68,9 @@ def test_l4casadi_deep_model_jac(self, deep_model):
 
         mx_inp = cs.MX.sym('x', deep_model.input_layer.in_features, 1)
 
-        jac_fun = cs.Function('f_jac', [mx_inp], [cs.jacobian(l4c.L4CasADi(deep_model, has_batch=True)(mx_inp), mx_inp)])
+        jac_fun = cs.Function('f_jac',
+                              [mx_inp],
+                              [cs.jacobian(l4c.L4CasADi(deep_model, model_expects_batch_dim=True)(mx_inp), mx_inp)])
 
         l4c_out = jac_fun(rand_inp.transpose(-2, -1).detach().numpy())