Spaces:
Sleeping
Sleeping
File size: 4,992 Bytes
0b7b562 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 |
import torch
def discriminator_loss(generator, discriminator, mol_graph, adj, annot, batch_size, device, grad_pen, lambda_gp,z_edge,z_node):
# Compute loss with real molecules.
logits_real_disc = discriminator(mol_graph)
prediction_real = - torch.mean(logits_real_disc)
# Compute loss with fake molecules.
node, edge, node_sample, edge_sample = generator(z_edge, z_node)
graph = torch.cat((node_sample.view(batch_size, -1), edge_sample.view(batch_size, -1)), dim=-1)
logits_fake_disc = discriminator(graph.detach())
prediction_fake = torch.mean(logits_fake_disc)
# Compute gradient loss.
eps = torch.rand(mol_graph.size(0),1).to(device)
x_int0 = (eps * mol_graph + (1. - eps) * graph).requires_grad_(True)
grad0 = discriminator(x_int0)
d_loss_gp = grad_pen(grad0, x_int0)
# Calculate total loss
d_loss = prediction_fake + prediction_real + d_loss_gp * lambda_gp
return node, edge,d_loss
def generator_loss(generator, discriminator, v, adj, annot, batch_size, penalty, matrices2mol, fps_r,submodel):
# Compute loss with fake molecules.
node, edge, node_sample, edge_sample = generator(adj, annot)
graph = torch.cat((node_sample.view(batch_size, -1), edge_sample.view(batch_size, -1)), dim=-1)
logits_fake_disc = discriminator(graph)
prediction_fake = - torch.mean(logits_fake_disc)
# Produce molecules.
g_edges_hat_sample = torch.max(edge_sample, -1)[1]
g_nodes_hat_sample = torch.max(node_sample , -1)[1]
fake_mol = [matrices2mol(n_.data.cpu().numpy(), e_.data.cpu().numpy(), strict=True)
for e_, n_ in zip(g_edges_hat_sample, g_nodes_hat_sample)]
g_loss = prediction_fake
# Compute penalty loss.
if submodel == "RL":
reward = penalty(fake_mol, fps_r)
# Reinforcement Loss
rew_fake = v(graph)
reward_loss = torch.mean(rew_fake) ** 2 + reward
# Calculate total loss
g_loss = prediction_fake + reward_loss * 1
return g_loss, fake_mol, g_edges_hat_sample, g_nodes_hat_sample, node, edge
def discriminator2_loss(generator, discriminator, mol_graph, adj, annot, batch_size, device, grad_pen, lambda_gp,akt1_adj,akt1_annot):
# Generate molecules.
dr_edges, dr_nodes = generator(adj,
annot,
akt1_adj,
akt1_annot)
dr_edges_hat = dr_edges.view(batch_size, -1)
dr_nodes_hat = dr_nodes.view(batch_size, -1)
dr_graph = torch.cat((dr_nodes_hat, dr_edges_hat), dim=-1)
# Compute loss with fake molecules.
dr_logits_fake = discriminator(dr_graph.detach())
d2_loss_fake = torch.mean(dr_logits_fake)
# Compute loss with real molecules.
dr_logits_real2 = discriminator(mol_graph)
d2_loss_real = - torch.mean(dr_logits_real2)
# Compute gradient loss.
eps_dr = torch.rand(mol_graph.size(0),1).to(device)
x_int0_dr = (eps_dr * mol_graph + (1. - eps_dr) * dr_graph).requires_grad_(True)
grad0_dr = discriminator(x_int0_dr)
d2_loss_gp = grad_pen(grad0_dr, x_int0_dr)
# Compute total loss.
d2_loss = d2_loss_fake + d2_loss_real + d2_loss_gp * lambda_gp
return d2_loss
def generator2_loss(generator, discriminator, v, adj, annot, batch_size, penalty, matrices2mol, fps_r,ak1_adj,akt1_annot, submodel):
# Generate molecules.
dr_edges_g, dr_nodes_g = generator(adj,
annot,
ak1_adj,
akt1_annot)
dr_edges_hat_g = dr_edges_g.view(batch_size, -1)
dr_nodes_hat_g = dr_nodes_g.view(batch_size, -1)
dr_graph_g = torch.cat((dr_nodes_hat_g, dr_edges_hat_g), dim=-1)
# Compute loss with fake molecules.
dr_g_edges_hat_sample, dr_g_nodes_hat_sample = torch.max(dr_edges_g, -1)[1], torch.max(dr_nodes_g, -1)[1]
g_tra_logits_fake2 = discriminator(dr_graph_g)
g2_loss_fake = - torch.mean(g_tra_logits_fake2)
# Reward
fake_mol_g = [matrices2mol(n_.data.cpu().numpy(), e_.data.cpu().numpy(), strict=True)
for e_, n_ in zip(dr_g_edges_hat_sample, dr_g_nodes_hat_sample)]
g2_loss = g2_loss_fake
if submodel == "RL":
reward2 = penalty(fake_mol_g, fps_r)
# Reinforcement Loss
rew_fake2 = v(dr_graph_g)
reward_loss2 = torch.mean(rew_fake2) ** 2 + reward2
# Calculate total loss
g2_loss = g2_loss_fake + reward_loss2 * 10
return g2_loss, fake_mol_g, dr_g_edges_hat_sample, dr_g_nodes_hat_sample#, reward2 |