Added MACL(Model Aware Contrastive learning ) Loss Function

[Abhishekjha3515]

Added MACL(Model Aware Contrastive learning ) Loss Function in Loss folder

[guarin]

Thanks for the contribution and sorry for the late review!

Could you add a minimal test in tests/loss/test_macl_loss.py? It is enough to pass some dummy values and check that the loss returns a valid value.

Could you also import the loss in lightly/loss/__init__.py? See https://github.com/lightly-ai/lightly/blob/master/lightly/loss/__init__.py, then it can be directly accessed as form lightly.loss import MACLLoss

[guarin]

Suggested change

	def __init__(self, temperature=0.1, alpha=0.5, A_0=0):
	super().__init__()
	self.t_0 = temperature
	self.alpha = alpha
	self.A_0 = A_0
	def __init__(self, temperature: float = 0.1, alpha: float = 0.5, A_0: float = 0):
	super().__init__()
	self.temperature = temperature
	self.alpha = alpha
	self.A_0 = A_0

Could you also add a short docstring that describes what the loss does, what the parameters are, and add a reference to the paper that introduced the loss? You can follow the example here:

lightly/lightly/loss/ntx_ent_loss.py

Line 17 in 356ae56

class NTXentLoss(nn.Module):

[Abhishekjha3515]

ok i will add the loss doc also