yzh119 · January 12, 2018 12:25 · Jan 12, 2018
diff --git a/st-gumbel.py b/st-gumbel.py
@@ -0,0 +1,30 @@
+from __future__ import print_function
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.autograd import Variable
+
+def sample_gumbel(shape, eps=1e-20):
+    U = torch.rand(shape).cuda()
+    return -Variable(torch.log(-torch.log(U + eps) + eps))
+
+def gumbel_softmax_sample(logits, temperature):
+    y = logits + sample_gumbel(logits.size())
+    return F.softmax(y / temperature, dim=-1)
+
+def gumbel_softmax(logits, temperature):
+    """
+    input: [*, n_class]
+    return: [*, n_class] an one-hot vector
+    """
+    y = gumbel_softmax_sample(logits, temperature)
+    shape = y.size()
+    _, ind = y.max(dim=-1)
+    y_hard = torch.zeros_like(y).view(-1, shape[-1])
+    y_hard.scatter_(1, ind.view(-1, 1), 1)
+    y_hard = y_hard.view(*shape)
+    return (y_hard - y).detach() + y
+
+if __name__ == '__main__':
+    import math
+    print(gumbel_softmax(Variable(torch.cuda.FloatTensor([[math.log(0.1), math.log(0.4), math.log(0.3), math.log(0.2)]] * 20000)),     0.8).sum(dim=0))