docs: update docs with binding name restrictions

Signed-off-by: Naren Dasan <[email protected]>
Signed-off-by: Naren Dasan <[email protected]>

Author: narendasan

cpp/include/torch_tensorrt/torch_tensorrt.h

@@ -739,7 +739,12 @@ TORCHTRT_API std::string convert_method_to_trt_engine(
  * module. Registers execution of the engine as the forward method of the module
  * Forward is defined as: forward(Tensor[]) -> Tensor[]
  *
- * @return: A new module trageting a TensorRT engine
+ * TensorRT bindings must have names with the following format:
+ * - [symbol].[index in input / output array]
+ * ex.
+ * - [x.0, x.1, x.2] -> [y.0]
+ *
+ * @return: A new module targeting a TensorRT engine
  */
 TORCHTRT_API torch::jit::Module embed_engine_in_new_module(const std::string& engine, Device device);
 } // namespace torchscript

py/torch_tensorrt/ts/_compiler.py

@@ -207,6 +207,11 @@ def embed_engine_in_new_module(serialized_engine: bytes, device=Device._current_
 
         forward(Tensor[]) -> Tensor[]
 
+    TensorRT bindings must have names with the following format:
+      - [symbol].[index in input / output array]
+      ex.
+      - [x.0, x.1, x.2] -> [y.0]
+
     Module can be save with engine embedded with torch.jit.save and moved / loaded according to torch_tensorrt portability rules
 
     Arguments: