lgatr.nets.conditional_lgatr_slim.ConditionalLGATrSlimBlock

class lgatr.nets.conditional_lgatr_slim.ConditionalLGATrSlimBlock(v_channels, condition_v_channels, s_channels, condition_s_channels, num_heads, nonlinearity='gelu', mlp_ratio=2, attn_ratio=1, num_layers_mlp=2, dropout_prob=None)[source]

Bases: Module

A single block of the conditional L-GATr-slim, consisting of self-attention, cross-attention and MLP layers, pre-norm and residual connections.

forward(vectors, vectors_condition, scalars, scalars_condition, attn_kwargs=None, crossattn_kwargs=None)[source]
Parameters:

  • vectors (torch.Tensor) – A tensor of shape (…, v_channels, 4) representing Lorentz vectors.

  • vectors_condition (torch.Tensor) – A tensor of shape (…, condition_v_channels, 4) representing a Lorentz vector condition included in cross-attention.

  • scalars (torch.Tensor) – A tensor of shape (…, s_channels) representing scalar features.

  • scalars_condition (torch.Tensor) – A tensor of shape (…, condition_s_channels) representing a scalar condition included in cross-attention.

  • **attn_kwargs (dict) – Additional keyword arguments for the attention function.

  • **crossattn_kwargs (dict) – Additional keyword arguments for the cross-attention function.

Returns:

Tensors of the same shape as input representing the normalized vectors and scalars.

Return type:

torch.Tensor, torch.Tensor