merge

blas.py

@@ -185,5 +185,5 @@ class GpuConv(Op):
                 logical_img_shape=self.logical_img_hw,
                 logical_kern_shape=self.logical_kern_hw,
                 kern_align=self.logical_kern_align_top,
-                verbose=1)
+                verbose=0)
 

nnet.py

@@ -224,8 +224,6 @@ class GpuCrossentropySoftmax1HotWithBiasDx (Op):
             }
         }
         {
-            std::cerr << "LAUNCHING NeW KEWNEL\\n";
-
             kCrossEntropySoftmax1HotWithBiasDx_%(nodename)s
                 <<<
                     CudaNdarray_HOST_DIMS(cnda_%(dx)s)[0],
@@ -233,6 +231,7 @@ class GpuCrossentropySoftmax1HotWithBiasDx (Op):
                 >>>(
                         CudaNdarray_HOST_DIMS(cnda_%(dx)s)[0],
                         CudaNdarray_HOST_DIMS(cnda_%(dx)s)[1], 
+
                         CudaNdarray_DEV_DATA(cnda_%(dnll)s),
                         CudaNdarray_HOST_STRIDES(cnda_%(dnll)s)[0],
 
@@ -265,23 +264,25 @@ class GpuCrossentropySoftmax1HotWithBiasDx (Op):
            const float * y_idx, const int y_idx_s0,
            float * dx)
         {
-        return;
-            for (size_t i = blockIdx.x; i < N; i += gridDim.x)
+            for (int i = blockIdx.x; i < N; i += gridDim.x)
             {
                 float dnll_i = dnll[i * dnll_s0];
                 int y_i = (int)y_idx[i * y_idx_s0];
 
-                for (size_t j = threadIdx.x; j < K; j += blockDim.x)
+                for (int j = threadIdx.x; j < K; j += blockDim.x)
                 {
                     if (y_i == j)
                     {
-                        dx[i * K + j] = dnll_i * (sm[i * sm_s0 + j * sm_s1]-1);
+                        dx[i * K + j] = dnll_i * (sm[i * sm_s0 + j * sm_s1]-1.0);
                     }
                     else
                     {
                         dx[i * K + j] = dnll_i * sm[i * sm_s0 + j * sm_s1];
                     }
+                    //dx[i * K + j] = dnll_i * sm[i * sm_s0 + j * sm_s1];
+                    //dx[i*K+j] = 0;
                 }
             }
         }
         """ % locals()
+

opt.py

@@ -255,7 +255,6 @@ def local_gpu_crossentorpy_softmax_argmax_1hot_with_bias(node):
 @register_opt()
 @local_optimizer([])
 def local_gpu_crossentorpy_softmax_1hot_with_bias_dx(node):
-    print 'REPLACING ', node, '??'
     if isinstance(node.op, tensor.nnet.CrossentropySoftmax1HotWithBiasDx):
         dnll,sm,yidx = node.inputs
         if sm.owner and sm.owner.op == host_from_gpu:
@@ -264,6 +263,5 @@ def local_gpu_crossentorpy_softmax_1hot_with_bias_dx(node):
                 gpu_from_host(dnll),
                 gpu_sm,
                 gpu_from_host(cast(yidx, 'float32')))
-            print 'YEP ', node
             return [host_from_gpu(gpu_dx)]
     return False

tests/test_nnet.py

@@ -221,7 +221,7 @@ def test_conv_nnet2():
         print rval_cpu[0], rval_gpu[0],rval_cpu[0]-rval_gpu[0]
         assert numpy.allclose(rval_cpu, rval_gpu,rtol=1e-4,atol=1e-4)
 
-def run_conv_nnet2_classif(shared_fn, isize, ksize):
+def run_conv_nnet2_classif(shared_fn, isize, ksize, n_iter=25):
 
     n_batch = 60
     shape_img = (n_batch, 1, isize, isize)
@@ -242,7 +242,7 @@ def run_conv_nnet2_classif(shared_fn, isize, ksize):
     b0 = shared_fn(numpy.asarray(numpy.zeros((n_kern,1,1)), dtype='float32'), 'b0')
     w1 = shared_fn(numpy.asarray(0.01*(numpy.random.rand(*shape_kern1)-0.5), dtype='float32'), 'w1')
     b1 = shared_fn(numpy.asarray(numpy.zeros((n_kern1,1,1)), dtype='float32'), 'b1')
-    v = shared_fn(numpy.asarray(numpy.zeros((n_hid, n_out)), dtype='float32'), 'c')
+    v = shared_fn(numpy.asarray(0.01*numpy.random.randn(n_hid, n_out), dtype='float32'), 'c')
     c = shared_fn(numpy.asarray(numpy.zeros(n_out), dtype='float32'), 'c')
 
     x = tensor.Tensor(dtype='float32', broadcastable=(0,0,0,0))('x')
@@ -267,15 +267,18 @@ def run_conv_nnet2_classif(shared_fn, isize, ksize):
     print 'building pfunc ...'
     train = pfunc([x,y,lr], [loss], mode=mode, updates=[(p, p-g) for p,g in zip(params, gparams)])
 
-    for i, n in enumerate(train.maker.env.toposort()):
-        print i, n
+    if theano.compile.mode.default_mode == 'PROFILE_MODE':
+        for i, n in enumerate(train.maker.env.toposort()):
+            print i, n
 
     xval = numpy.asarray(numpy.random.rand(*shape_img), dtype='float32')
-    yval = numpy.asarray(numpy.random.rand(n_batch,n_out), dtype='int32')#FRED: THIS DON'T WORK. THIS SET YVAL TO ALL ZERO!
-    lr = numpy.asarray(0.01, dtype='float32')
+    yval = numpy.asarray(numpy.random.rand(n_batch,n_out), dtype='float32')
+    lr = numpy.asarray(0.001, dtype='float32')
 
-    for i in xrange(10):
+    for i in xrange(n_iter):
         rval = train(xval, yval, lr)
+        if i % 10 == 0:
+            print 'rval', rval
     print_mode(mode)
     return rval