fix loss function labels

simclr.py

@@ -34,19 +34,38 @@ class SimCLR(object):
         self.criterion = torch.nn.CrossEntropyLoss().to(self.args.device)
 
     def info_nce_loss(self, features):
-        batch_targets = torch.arange(self.args.batch_size, dtype=torch.long).to(self.args.device)
-        batch_targets = torch.cat(self.args.n_views * [batch_targets])
+
+        labels = torch.cat([torch.arange(self.args.batch_size) for i in range(self.args.n_views)], dim=0)
+        labels = (labels.unsqueeze(0) == labels.unsqueeze(1)).float()
+        labels = labels.to(self.args.device)
 
         features = F.normalize(features, dim=1)
 
         similarity_matrix = torch.matmul(features, features.T)
         # assert similarity_matrix.shape == (
         #     self.args.n_views * self.args.batch_size, self.args.n_views * self.args.batch_size)
+        # assert similarity_matrix.shape == labels.shape
 
-        mask = torch.eye(len(batch_targets)).to(self.args.device)
-        similarities = similarity_matrix[~mask.bool()].view(similarity_matrix.shape[0], -1)
-        similarities = similarities / self.args.temperature
-        return similarities, batch_targets
+        # discard the main diagonal from both: labels and similarities matrix
+        mask = torch.eye(labels.shape[0], dtype=torch.bool).to(self.args.device)
+        labels = labels[~mask].view(labels.shape[0], -1)
+        similarity_matrix = similarity_matrix[~mask].view(similarity_matrix.shape[0], -1)
+        # assert similarity_matrix.shape == labels.shape
+
+        # select and combine multiple positives
+        positives = similarity_matrix[labels.bool()].view(labels.shape[0], -1)
+
+        # if there is more than one potive (n_views >= 2) combine the multiple positives
+        positives = positives.mean(dim=1).unsqueeze(1)
+
+        # select only the negatives the negatives
+        negatives = similarity_matrix[~labels.bool()].view(similarity_matrix.shape[0], -1)
+
+        logits = torch.cat([positives, negatives], dim=1)
+        labels = torch.zeros(logits.shape[0]).to(self.args.device)
+
+        logits = logits / self.args.temperature
+        return logits, labels
 
     def train(self, train_loader):