BCELoss add weight

python/jittor/nn.py

@@ -167,11 +167,16 @@ def cross_entropy_loss(output, target, ignore_index=None):
 def mse_loss(output, target):
     return (output-target).sqr().mean()
 
-def bce_loss(output, target, size_average=True):
+def bce_loss(output, target, weight=None, size_average=True):
+    loss = - (target * jt.log(jt.maximum(output, 1e-20)) + (1 - target) * jt.log(jt.maximum(1 - output, 1e-20)))
+
+    if weight is not None:
+        loss *= weight
+    
     if size_average:
-        return - (target * jt.log(jt.maximum(output, 1e-20)) + (1 - target) * jt.log(jt.maximum(1 - output, 1e-20))).mean()
+        return loss.mean()
     else:
-        return - (target * jt.log(jt.maximum(output, 1e-20)) + (1 - target) * jt.log(jt.maximum(1 - output, 1e-20))).sum()
+        return loss.sum()
 
 def l1_loss(output, target):
     return (output-target).abs().mean()
@@ -189,10 +194,11 @@ class MSELoss(Module):
         return mse_loss(output, target)
 
 class BCELoss(Module):
-    def __init__(self):
+    def __init__(self, weight=None, size_average=True):
-        pass
+        self.weight = weight
-    def execute(self, output, target, size_average=True):
+        self.size_average = size_average
-        return bce_loss(output, target, size_average)
+    def execute(self, output, target):
+        return bce_loss(output, target, self.weight, self.size_average)
 
 class L1Loss(Module):
     def __init__(self):
@@ -201,14 +207,17 @@ class L1Loss(Module):
         return l1_loss(output, target)
 
 class BCEWithLogitsLoss(Module):
-    def __init__(self):
+    def __init__(self, weight=None, size_average=True):
         self.sigmoid = Sigmoid()
-        self.bce = BCELoss()
+        self.bce = BCELoss(weight, size_average)
-    def execute(self, output, target, size_average=True):
+    def execute(self, output, target):
         output = self.sigmoid(output)
-        output = self.bce(output, target, size_average)
+        output = self.bce(output, target)
         return output
 
+def binary_cross_entropy_with_logits(input, target, weight=None, size_average=True):
+    return BCEWithLogitsLoss(weight, size_average)(input, target)
+
 def softmax(x, dim = None):
     if dim is None:
         x = (x - x.max()).exp()

python/jittor/test/test_loss.py

@@ -49,7 +49,7 @@ class TestLoss(unittest.TestCase):
         jt_y=jt_loss(jt.array(output), jt.array(target))
         tc_y=tc_loss(torch.from_numpy(output), torch.from_numpy(target))
         assert np.allclose(jt_y.numpy(), tc_y.numpy())
-        
+
     def test_bce_loss(self):
         jt_loss=jnn.BCELoss()
         tc_loss=tnn.BCELoss()
@@ -60,6 +60,13 @@ class TestLoss(unittest.TestCase):
         jt_y=jt_loss(jt_sig(jt.array(output)), jt.array(target))
         tc_y=tc_loss(tc_sig(torch.from_numpy(output)), torch.from_numpy(target))
         assert np.allclose(jt_y.numpy(), tc_y.numpy())
+
+        weight=np.random.randn(100).astype(np.float32)
+        jt_loss=jnn.BCELoss(weight=jt.array(weight), size_average=False)
+        tc_loss=tnn.BCELoss(weight=torch.Tensor(weight), size_average=False)
+        jt_y=jt_loss(jt_sig(jt.array(output)), jt.array(target))
+        tc_y=tc_loss(tc_sig(torch.from_numpy(output)), torch.from_numpy(target))
+        assert np.allclose(jt_y.numpy(), tc_y.numpy())
         
     def test_bce_with_logits_loss(self):
         jt_loss=jnn.BCEWithLogitsLoss()

python/jittor/utils/pytorch_converter.py

@@ -78,6 +78,32 @@ pjmap = {
         'extras': {},
         'delete': ['inplace'],
     },
+    'relu': {
+        'pytorch': {
+            'args': 'input', 
+        },
+        'jittor': {
+            'module': 'nn',
+            'name': 'relu',
+            'args': 'x'
+        },
+        'links': {'input': 'x'},
+        'extras': {},
+        'delete': [],
+    },
+    'binary_cross_entropy_with_logits': {
+        'pytorch': {
+            'args': 'input, target, weight, size_average=True', 
+        },
+        'jittor': {
+            'module': 'nn',
+            'name': 'binary_cross_entropy_with_logits',
+            'args': 'input, target, weight, size_average=True'
+        },
+        'links': {},
+        'extras': {},
+        'delete': [],
+    },
     'ReLU6': {
         'pytorch': {
             'args': 'inplace=False', 
@@ -274,6 +300,23 @@ pjmap = {
         'links': {},
         'extras': {},
     },
+    'clamp': {
+        'pytorch': {
+            'prefix': ['torch'],
+            'args_prefix': 'input, min, max, out=None',
+            'args': 'min, max, out=None',
+        },
+        'jittor': {
+            'prefix': 'jt',
+            'module': '',
+            'name': 'clamp',
+            'args_prefix': 'x, min_v, max_v',
+            'args': 'min_v, max_v'
+        },
+        'links': {'min': 'min_v', 'max': 'max_v'},
+        'extras': {},
+        'delete': ['out'],
+    },
     'permute': {
         'pytorch': {
             'prefix': [],
@@ -354,7 +397,7 @@ unsupport_ops = [
     # ***************************************************************
     # torch.nn
     # ***************************************************************
-    'Parameter', 'ModuleList', 'ModuleDict', 'ParameterList', 'ParameterDict', 
+    'Parameter', 'ModuleDict', 'ParameterList', 'ParameterDict', 
     'Conv1d', 'Conv3d', 'ConvTranspose1d', 'ConvTranspose3d', 'Unfold', 'Fold', 
     'MaxPool1d', 'MaxPool3d', 'MaxUnpool1d', 'MaxUnpool2d', 'MaxUnpool3d', 'AvgPool1d', 
     'AvgPool3d', 'FractionalMaxPool2d', 'LPPool1d', 'LPPool2d', 'AdaptiveMaxPool1d',