[RELAY/OP] Gradient of relay level1 ops (apache#2633)

python/tvm/relay/expr.py

@@ -51,6 +51,9 @@ def astype(self, dtype):
         """
         return _make.cast(self, dtype)
 
+    def __neg__(self):
+        return _op_make.negative(self)
+
     def __add__(self, other):
         if isinstance(other, Expr):
             return _op_make.add(self, other)

python/tvm/relay/op/__init__.py

@@ -18,6 +18,7 @@
 
 # operator registry
 from . import _tensor
+from . import _tensor_grad
 from . import _transform
 from . import _reduce
 from ..expr import Expr

python/tvm/relay/op/_tensor.py

@@ -3,25 +3,7 @@
 from __future__ import absolute_import
 import topi
 from .op import register_compute, register_schedule, register_pattern
-from .op import register_gradient
 from .op import schedule_injective, OpPattern
-from .transform import collapse_sum_like
-from .tensor import negative
-
-
-def add_grad(orig, grad):
-    return [collapse_sum_like(grad, orig.args[0]), collapse_sum_like(grad, orig.args[1])]
-
-
-register_gradient("add", add_grad)
-
-
-def subtract_grad(orig, grad):
-    return [collapse_sum_like(grad, orig.args[0]),
-            collapse_sum_like(negative(grad), orig.args[1])]
-
-
-register_gradient("subtract", subtract_grad)
 
 schedule_broadcast = schedule_injective
 schedule_elemwise = schedule_injective

python/tvm/relay/op/_tensor_grad.py

@@ -0,0 +1,79 @@
+#pylint: disable=invalid-name, unused-argument
+"""Backend compiler related feature registration"""
+from __future__ import absolute_import
+from ..expr import const
+from .op import register_gradient
+from .transform import collapse_sum_like, where
+from .tensor import exp, negative, power, less
+from .tensor import zeros_like, ones_like
+
+
+@register_gradient("log")
+def log_grad(orig, grad):
+    """Returns [grad * (1 / x)]"""
+    x = orig.args[0]
+    return [grad * ones_like(x) / x]
+
+
+@register_gradient("exp")
+def exp_grad(orig, grad):
+    """Returns [grad * exp(x)]"""
+    return [grad * exp(orig.args[0])]
+
+
+@register_gradient("sqrt")
+def sqrt_grad(orig, grad):
+    """Returns [grad * 0.5 * (x ^ -0.5)]"""
+    a = const(0.5)  # (TODO) type?
+    return [grad * a * power(orig.args[0], negative(a))]
+
+
+@register_gradient("sigmoid")
+def sigmoid_grad(orig, grad):
+    """Returns [grad * sigmoid(x) * (1 - sigmoid(x))]."""
+    return [grad * orig * (ones_like(orig) - orig)]
+
+
+@register_gradient("tanh")
+def tanh_grad(orig, grad):
+    """Returns grad * (1 - tanh(x) * tanh(x))."""
+    return [grad * ones_like(orig) - orig * orig]
+
+
+@register_gradient("nn.relu")
+def relu_grad(orig, grad):
+    """Returns grad * (select(x < 0, 0, 1))."""
+    x = orig.args[0]
+    zeros = zeros_like(x)
+    ones = ones_like(x)
+    return [where(less(x, zeros), zeros, ones * grad)]
+
+
+@register_gradient("add")
+def add_grad(orig, grad):
+    """Returns [grad, grad]"""
+    return [collapse_sum_like(grad, orig.args[0]),
+            collapse_sum_like(grad, orig.args[1])]
+
+
+@register_gradient("subtract")
+def subtract_grad(orig, grad):
+    """Returns [grad, -grad]"""
+    return [collapse_sum_like(grad, orig.args[0]),
+            collapse_sum_like(negative(grad), orig.args[1])]
+
+
+@register_gradient("multiply")
+def multiply_grad(orig, grad):
+    """Returns [grad * y, grad * x]"""
+    x, y = orig.args
+    return [collapse_sum_like(grad * y, x),
+            collapse_sum_like(grad * x, y)]
+
+
+@register_gradient("divide")
+def divide_grad(orig, grad):
+    """Returns [grad / y,  - grad * (x / y) / y]"""
+    x, y = orig.args
+    return [collapse_sum_like(grad / y, x),
+            collapse_sum_like(- (grad * orig / y), y)]

python/tvm/relay/op/op.py

@@ -168,7 +168,7 @@ def register_pattern(op_name, pattern, level=10):
     """
     return register(op_name, "TOpPattern", pattern, level)
 
-def register_gradient(op_name, fgradient, level=10):
+def register_gradient(op_name, fgradient=None, level=10):
     """Register operator pattern for an op.
 
     Parameters

tests/python/relay/test_op_grad_level1.py

@@ -0,0 +1,76 @@
+import tvm
+import numpy as np
+from tvm import relay
+from tvm.relay.ir_pass import gradient, infer_type
+from tvm.relay.testing import ctx_list
+
+def sigmoid(x):
+    one = np.ones_like(x)
+    return one / (one + np.exp(-x))
+
+def relu(x):
+    x_copy = np.copy(x)
+    np.maximum(x_copy, 0, x_copy)
+    return x_copy
+
+def test_unary_op():
+    def check_single_op(opfunc, ref):
+        shape = (10, 4)
+        dtype = 'float32'
+        tp = relay.TensorType(shape, dtype)
+        x = relay.var("x", tp)
+        y = opfunc(x)
+
+        if ref is not None:
+            data = np.random.rand(*shape).astype(dtype)
+            ref_grad = ref(data)
+            fwd_func = relay.Function([x], y)
+            bwd_func = infer_type(gradient(fwd_func))
+
+            for target, ctx in ctx_list():
+                intrp = relay.create_executor(ctx=ctx, target=target)
+                op_res, (op_grad, ) = intrp.evaluate(bwd_func)(data)
+                np.testing.assert_allclose(op_grad.asnumpy(), ref_grad, rtol=0.01)
+
+    for opfunc, ref in [(tvm.relay.log, lambda x: 1 / x),
+                        (tvm.relay.exp, np.exp),
+                        (tvm.relay.sigmoid, lambda x: sigmoid(x) * (1 - sigmoid(x))),
+                        (tvm.relay.tanh, lambda x: 1 - np.tanh(x) * np.tanh(x)),
+                        (tvm.relay.sqrt, lambda x: 0.5 * np.power(x, -0.5)),
+                        (relay.nn.relu, lambda x: np.where(x < 0, np.zeros_like(x), np.ones_like(x)))]:
+        check_single_op(opfunc, ref)
+
+
+def test_binary_op():
+    def inst(vars, sh):
+        return [vars.get(s, s) for s in sh]
+
+    def check_binary_op(opfunc, ref):
+        s = (5, 10, 5)
+        t = relay.TensorType((5, 10, 5))
+        x = relay.var("x", t)
+        y = relay.var("y", t)
+        z = opfunc(x, y)
+
+        x_data = np.random.rand(*s).astype(t.dtype)
+        y_data = np.random.rand(*s).astype(t.dtype)
+        ref_grad0, ref_grad1 = ref(x_data, y_data)
+        fwd_func = relay.Function([x, y], z)
+        bwd_func = infer_type(gradient(fwd_func))
+
+        for target, ctx in ctx_list():
+            intrp = relay.create_executor(ctx=ctx, target=target)
+            op_res, (op_grad0, op_grad1) = intrp.evaluate(bwd_func)(x_data, y_data)
+            np.testing.assert_allclose(op_grad0.asnumpy(), ref_grad0, rtol=0.01)
+            np.testing.assert_allclose(op_grad1.asnumpy(), ref_grad1, rtol=0.01)
+
+    for opfunc, ref in [(relay.add, lambda x, y: [np.ones_like(x), np.ones_like(y)]),
+                        (relay.subtract, lambda x, y: [np.ones_like(x), -np.ones_like(y)]),
+                        (relay.multiply, lambda x, y: [y, x]),
+                        (relay.divide, lambda x, y: [1 / y, - x / (y**2)])]:
+        check_binary_op(opfunc, ref)
+
+
+if __name__ == "__main__":
+    test_unary_op()
+    test_binary_op()

tests/python/relay/test_op_level1.py

@@ -39,11 +39,11 @@ def check_single_op(opfunc, ref):
 
 
     for opfunc, ref in [(tvm.relay.log, np.log),
-                   (tvm.relay.exp, np.exp),
-                   (tvm.relay.sqrt, np.sqrt),
-                   (tvm.relay.sigmoid, sigmoid),
-                   (tvm.relay.tanh, np.tanh),
-                   (relay.nn.relu, relu)]:
+                        (tvm.relay.exp, np.exp),
+                        (tvm.relay.sqrt, np.sqrt),
+                        (tvm.relay.sigmoid, sigmoid),
+                        (tvm.relay.tanh, np.tanh),
+                        (relay.nn.relu, relu)]:
         check_single_op(opfunc, ref)
 
 
@@ -84,9 +84,9 @@ def check_binary_op(opfunc, ref):
                 np.testing.assert_allclose(op_res.asnumpy(), ref_res, rtol=0.01)
 
     for opfunc, ref in [(relay.add, np.add),
-                   (relay.subtract, np.subtract),
-                   (relay.multiply, np.multiply),
-                   (relay.divide, np.divide)]:
+                        (relay.subtract, np.subtract),
+                        (relay.multiply, np.multiply),
+                        (relay.divide, np.divide)]:
         check_binary_op(opfunc, ref)