Based on C$7\_2,$ time$-$dependent heat transfer, perform the following tasks:

In a PINN, the loss can become imbalanced between different terms $($boundary conditions, initial conditions, and PDE residual$).$ You can manually balance these loss terms by adding weights to each loss term:

loss $=w_{-}bc*los{s}_{b}c+w_{i}c*los{s}_{i}c+w_{p}de*los{s}_{p}de$

Adjust the weights $w_{-}bc,w_{-}c,w_{-}$pde to balance the contributions to accelerate training.

$2.$ Try changing the activation function. In PINNs, tanh or sin activations often work better for smooth solutions than the commonly used ReLU. Compare the effects of tanh$0,$ sin$0,$ and Reluo

Please provide code and results! thank you