diff --git a/senteval/tools/relatedness.py b/senteval/tools/relatedness.py
index d8bacce5..25d8dc1a 100644
--- a/senteval/tools/relatedness.py
+++ b/senteval/tools/relatedness.py
@@ -19,6 +19,7 @@
 
 from scipy.stats import pearsonr
 
+device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
 
 class RelatednessPytorch(object):
     # Can be used for SICK-Relatedness, and STS14
@@ -26,8 +27,8 @@ def __init__(self, train, valid, test, devscores, config):
         # fix seed
         np.random.seed(config['seed'])
         torch.manual_seed(config['seed'])
-        assert torch.cuda.is_available(), 'torch.cuda required for Relatedness'
-        torch.cuda.manual_seed(config['seed'])
+        if torch.cuda.is_available():
+            torch.cuda.manual_seed(config['seed'])
 
         self.train = train
         self.valid = valid
@@ -58,12 +59,12 @@ def __init__(self, train, valid, test, devscores, config):
 
     def prepare_data(self, trainX, trainy, devX, devy, testX, testy):
         # Transform probs to log-probs for KL-divergence
-        trainX = torch.from_numpy(trainX).float().cuda()
-        trainy = torch.from_numpy(trainy).float().cuda()
-        devX = torch.from_numpy(devX).float().cuda()
-        devy = torch.from_numpy(devy).float().cuda()
-        testX = torch.from_numpy(testX).float().cuda()
-        testY = torch.from_numpy(testy).float().cuda()
+        trainX = torch.from_numpy(trainX).float().to(device)
+        trainy = torch.from_numpy(trainy).float().to(device)
+        devX = torch.from_numpy(devX).float().to(device)
+        devy = torch.from_numpy(devy).float().to(device)
+        testX = torch.from_numpy(testX).float().to(device)
+        testY = torch.from_numpy(testy).float().to(device)
 
         return trainX, trainy, devX, devy, testX, testy
 
@@ -107,7 +108,7 @@ def trainepoch(self, X, y, nepoches=1):
             all_costs = []
             for i in range(0, len(X), self.batch_size):
                 # forward
-                idx = torch.from_numpy(permutation[i:i + self.batch_size]).long().cuda()
+                idx = torch.from_numpy(permutation[i:i + self.batch_size]).long().to(device)
                 Xbatch = X[idx]
                 ybatch = y[idx]
                 output = self.model(Xbatch)