test_pass_fold_scale_axis.py 23.1 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.
17
import numpy as np
18

Zhi committed
19 20 21
from tvm import relay
from tvm.relay import transform

22 23 24
def _get_positive_scale(size):
    return np.random.uniform(0.5, 1, size=size).astype('float32')

25

Zhi committed
26 27 28 29
def run_opt_pass(expr, opt_pass):
    assert isinstance(opt_pass, transform.Pass)
    mod = relay.Module.from_expr(expr)
    mod = opt_pass(mod)
30
    entry = mod["main"]
Zhi committed
31 32 33
    return entry if isinstance(expr, relay.Function) else entry.body


34 35 36
def test_fold_fwd_simple():
    """Simple testcase."""
    def before(x, conv_weight, in_bias, in_scale, channels):
37
        args = [x, conv_weight, in_bias]
38 39 40 41 42 43 44 45
        in_bias = relay.expand_dims(in_bias, axis=1, num_newaxis=2)
        x = relay.multiply(x, in_scale)
        x = relay.nn.relu(x)
        x = relay.add(x, in_bias)
        y = relay.nn.conv2d(x, conv_weight,
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
46

47 48 49 50
        return relay.Function(args, y)

    def expected(x, conv_weight, in_bias, in_scale, channels):
        # use a fixed order of args so alpha equal check can pass
51
        args = [x, conv_weight, in_bias]
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69
        in_bias = relay.expand_dims(in_bias, axis=1, num_newaxis=2)
        squeezed_scale = relay.squeeze(in_scale, axis=[1,2])
        x = relay.nn.relu(x)
        in_bias = relay.divide(in_bias, relay.expand_dims(squeezed_scale, axis=1, num_newaxis=2))
        x = relay.add(x, in_bias)
        conv_weight = relay.multiply(
            conv_weight , relay.expand_dims(squeezed_scale, axis=1, num_newaxis=2))
        y = relay.nn.conv2d(x, conv_weight,
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        return relay.Function(args, y)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        weight = relay.var("weight")
        in_bias = relay.var("in_bias", shape=(in_channels,))
70
        in_scale = relay.const(_get_positive_scale((in_channels, 1, 1)))
71
        y1 = before(x, weight, in_bias, in_scale, channels)
Zhi committed
72
        y1 = run_opt_pass(y1, transform.InferType())
73 74
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
Zhi committed
75
        y1_folded = run_opt_pass(y1, transform.ForwardFoldScaleAxis())
76
        y1_expected = expected(x, weight, in_bias, in_scale, channels)
77

Zhi committed
78 79 80
        y1_folded = run_opt_pass(y1_folded, transform.InferType())
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
81 82 83 84 85 86 87

    check((2, 4, 10, 10), 2)


def test_fold_fwd_dual_path():
    """scale axis being consumed by two consumers"""
    def before(x, conv_weight, in_bias, in_scale, channels):
88
        args = [x, conv_weight, in_bias]
89 90 91 92 93 94 95
        x = relay.multiply(in_scale, x)
        x = relay.nn.relu(x)
        x = relay.subtract(x, in_bias)
        y1 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NHWC",
96
                             kernel_layout="HWIO",
97 98 99 100 101 102
                             groups=channels,
                             padding=(1, 1))
        y2 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NHWC",
103
                             kernel_layout="HWIO",
104 105 106 107 108 109
                             groups=channels,
                             padding=(1, 1))
        z = relay.add(y1, y2)
        return relay.Function(args, z)

    def expected(x, conv_weight, in_bias, in_scale, channels):
110
        args = [x, conv_weight, in_bias]
111 112 113 114 115 116 117 118
        x = relay.nn.relu(x)
        in_bias = relay.divide(in_bias, in_scale)
        x = relay.subtract(x, in_bias)
        y1 = relay.nn.conv2d(x,
                             relay.multiply(conv_weight, in_scale),
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NHWC",
119
                             kernel_layout="HWIO",
120 121 122 123 124 125 126
                             groups=channels,
                             padding=(1, 1))
        y2 = relay.nn.conv2d(x,
                             relay.multiply(conv_weight, in_scale),
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NHWC",
127
                             kernel_layout="HWIO",
128 129 130 131 132 133 134 135 136 137 138 139
                             groups=channels,
                             padding=(1, 1))
        z = relay.add(y1, y2)
        return relay.Function(args, z)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[-1]
        # test depthwise
        assert in_channels == channels
        weight = relay.var("weight")
        in_bias = relay.var("in_bias", shape=(in_channels,))
140
        in_scale = relay.const(_get_positive_scale(in_channels,))
141
        y1 = before(x, weight, in_bias, in_scale, channels)
Zhi committed
142 143
        y1 = run_opt_pass(y1, transform.InferType())
        y1_folded = run_opt_pass(y1, transform.ForwardFoldScaleAxis())
144 145 146
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
        y1_expected = expected(x, weight, in_bias, in_scale, channels)
Zhi committed
147 148
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163

    check((2, 4, 10, 3), 3)


def test_fold_fwd_fail():
    """testcase where we canont fold"""
    def before(x, conv_weight, in_bias, in_scale, channels):
        x = relay.multiply(x, in_scale)
        xx = relay.nn.leaky_relu(x, alpha=0.1)
        y1 = relay.nn.conv2d(xx, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NHWC",
                             padding=(1, 1))
        z = relay.add(y1, x)
Zhi committed
164
        return relay.Function(relay.analysis.free_vars(z), z)
165 166 167 168 169 170 171 172

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[-1]
        # test depthwise
        assert in_channels == channels
        weight = relay.var("weight")
        in_bias = relay.var("in_bias", shape=(in_channels,))
173
        in_scale = relay.const(_get_positive_scale(size=(in_channels,)))
174
        y1 = before(x, weight, in_bias, in_scale, channels)
Zhi committed
175 176 177
        y1 = run_opt_pass(y1, transform.InferType())
        y1_folded = run_opt_pass(y1, transform.ForwardFoldScaleAxis())
        assert relay.analysis.alpha_equal(y1, y1_folded)
178 179 180 181

    check((2, 11, 10, 4), 4)


182 183 184 185 186 187 188 189 190 191 192
def test_fold_fwd_relu_fail():
    """testcase where we canont fold because scale can not pass relu"""
    def before(x, conv_weight, in_bias, in_scale, channels):
        x = relay.multiply(x, in_scale)
        xx = relay.nn.relu(x)
        y1 = relay.nn.conv2d(xx, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NHWC",
                             padding=(1, 1))
        z = relay.add(y1, x)
Zhi committed
193
        return relay.Function(relay.analysis.free_vars(z), z)
194 195 196 197 198 199 200 201 202

    def check(shape, channels, in_scale):
        x =  relay.var("x", shape=shape)
        in_channels = shape[-1]
        # test depthwise
        assert in_channels == channels
        weight = relay.var("weight")
        in_bias = relay.var("in_bias", shape=(in_channels,))
        y1 = before(x, weight, in_bias, in_scale, channels)
Zhi committed
203 204 205
        y1 = run_opt_pass(y1, transform.InferType())
        y1_folded = run_opt_pass(y1, transform.ForwardFoldScaleAxis())
        assert relay.analysis.alpha_equal(y1, y1_folded)
206 207 208

    in_scale = relay.var("in_scale", shape=(4,))
    check((2, 11, 10, 4), 4, in_scale)
209
    in_scale = relay.const(-_get_positive_scale((4,)))
210 211 212
    check((2, 11, 10, 4), 4, in_scale)


213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
def test_fold_fwd_negative_scale():
    """Testcase of folding negative scale"""
    def before(x, conv_weight, in_scale, channels):
        args = [x, conv_weight]
        x = relay.multiply(x, in_scale)
        y = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        return relay.Function(args, y)

    def expected(x, conv_weight, in_scale, channels):
        # use a fixed order of args so alpha equal check can pass
        args = [x, conv_weight]
        squeezed_scale = relay.squeeze(in_scale, axis=[1,2])
        conv_weight = relay.multiply(
            conv_weight , relay.expand_dims(squeezed_scale, axis=1, num_newaxis=2))
        y = relay.nn.conv2d(x,
                             conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        return relay.Function(args, y)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        in_scale = relay.const(-_get_positive_scale((in_channels, 1, 1)))
        weight = relay.var("weight")
        y1 = before(x, weight, in_scale, channels)
Zhi committed
243
        y1 = run_opt_pass(y1, transform.InferType())
244 245
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
Zhi committed
246
        y1_folded = run_opt_pass(y1, transform.ForwardFoldScaleAxis())
247
        y1_expected = expected(x, weight, in_scale, channels)
Zhi committed
248 249
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
250 251 252 253

    check((2, 4, 10, 10), 4)


254 255 256
def test_fold_bwd_simple():
    """Simple testcase."""
    def before(x, conv_weight, out_bias, out_scale, channels):
257
        args = [x, conv_weight, out_bias]
258 259 260 261 262 263 264 265 266 267 268 269
        out_bias = relay.expand_dims(out_bias, axis=1, num_newaxis=2)
        y = relay.nn.conv2d(x, conv_weight,
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y = relay.add(y, out_bias)
        y = relay.nn.relu(y)
        y = relay.multiply(y, out_scale)
        return relay.Function(args, y)

    def expected(x, conv_weight, out_bias, out_scale, channels):
        # use a fixed order of args so alpha equal check can pass
270
        args = [x, conv_weight, out_bias]
271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
        out_bias = relay.expand_dims(out_bias, axis=1, num_newaxis=2)
        squeezed_scale = relay.squeeze(out_scale, axis=[1,2])
        conv_weight = relay.multiply(
            conv_weight , relay.expand_dims(squeezed_scale, axis=1, num_newaxis=3))

        y = relay.nn.conv2d(x, conv_weight,
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        out_bias = relay.multiply(out_bias,
                                  relay.expand_dims(squeezed_scale, axis=1, num_newaxis=2))
        y = relay.add(y, out_bias)
        y = relay.nn.relu(y)
        return relay.Function(args, y)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        weight = relay.var("weight")
        out_bias = relay.var("out_bias", shape=(channels,))
291
        out_scale = relay.const(_get_positive_scale((channels, 1, 1)))
292 293

        y1 = before(x, weight, out_bias, out_scale, channels)
Zhi committed
294
        y1 = run_opt_pass(y1, transform.InferType())
295 296
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
Zhi committed
297
        y1_folded = run_opt_pass(y1, transform.BackwardFoldScaleAxis())
298
        y1_expected = expected(x, weight, out_bias, out_scale, channels)
Zhi committed
299 300
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
301 302 303 304 305 306 307

    check((2, 4, 10, 10), 8)


def test_fold_bwd_dual_path():
    """Dual path testcase."""
    def before(x, conv_weight, out_bias, out_scale, channels):
308
        args = [x, conv_weight, out_bias]
309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324
        y1 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y1 = relay.nn.relu(y1)
        y2 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y2 = relay.nn.relu(y2)
        y = relay.add(y1, y2)
        y = relay.multiply(y, out_scale)
        return relay.Function(args, y)

    def expected(x, conv_weight, out_bias, out_scale, channels):
        # use a fixed order of args so alpha equal check can pass
325
        args = [x, conv_weight, out_bias]
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
        out_bias = relay.expand_dims(out_bias, axis=1, num_newaxis=2)
        squeezed_scale = relay.squeeze(out_scale, axis=[1,2])
        def fold_conv_weight():
            return  relay.multiply(
                conv_weight ,
                relay.expand_dims(squeezed_scale, axis=1, num_newaxis=3))
        y1 = relay.nn.conv2d(x, fold_conv_weight(),
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y1 = relay.nn.relu(y1)
        y2 = relay.nn.conv2d(x, fold_conv_weight(),
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y2 = relay.nn.relu(y2)
        y = relay.add(y1, y2)
        return relay.Function(args, y)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        weight = relay.var("weight")
        out_bias = relay.var("out_bias", shape=(channels,))
350
        out_scale = relay.const(_get_positive_scale((channels, 1, 1)))
351 352

        y1 = before(x, weight, out_bias, out_scale, channels)
Zhi committed
353
        y1 = run_opt_pass(y1, transform.InferType())
354 355
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
Zhi committed
356
        y1_folded = run_opt_pass(y1, transform.BackwardFoldScaleAxis())
357
        y1_expected = expected(x, weight, out_bias, out_scale, channels)
Zhi committed
358 359
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
360 361 362 363

    check((2, 4, 10, 10), 8)


364 365
def test_fold_bwd_dual_consumer():
    def before(x, conv_weight, out_bias, out_scale, channels):
366
        args = [x, conv_weight, out_bias]
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
        y0 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y0 = relay.multiply(y0, out_scale)
        y0 = relay.nn.relu(y0)

        y1 = relay.nn.conv2d(y0, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y1 = relay.multiply(y1, out_scale)
        y1 = relay.nn.relu(y1)

        y2 = relay.nn.conv2d(y0, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y2 = relay.multiply(y2, out_scale)
        y2 = relay.nn.relu(y2)

        y = relay.add(y1, y2)
        return relay.Function(args, y)

    def expected(x, conv_weight, out_bias, out_scale, channels):
        # use a fixed order of args so alpha equal check can pass
393
        args = [x, conv_weight, out_bias]
394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
        def fold_conv_weight():
            squeezed_scale = relay.squeeze(out_scale, axis=[1,2])
            return  relay.multiply(
                conv_weight ,
                relay.expand_dims(squeezed_scale, axis=1, num_newaxis=3))
        y0 = relay.nn.conv2d(x, fold_conv_weight(),
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y0 = relay.nn.relu(y0)
        y1 = relay.nn.conv2d(y0, fold_conv_weight(),
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y1 = relay.nn.relu(y1)
        y2 = relay.nn.conv2d(y0, fold_conv_weight(),
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y2 = relay.nn.relu(y2)
        y = relay.add(y1, y2)
        return relay.Function(args, y)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        weight = relay.var("weight")
        out_bias = relay.var("out_bias", shape=(channels,))
422
        out_scale = relay.const(_get_positive_scale((channels,1, 1)))
423 424

        y1 = before(x, weight, out_bias, out_scale, channels)
Zhi committed
425
        y1 = run_opt_pass(y1, transform.InferType())
426 427
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
Zhi committed
428
        y1_folded = run_opt_pass(y1, transform.BackwardFoldScaleAxis())
429
        y1_expected = expected(x, weight, out_bias, out_scale, channels)
Zhi committed
430 431
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
432 433 434 435

    check((2, 4, 10, 10), 4)


436 437 438
def test_fold_bwd_fail():
    """Dual path testcase."""
    def fail1(x, conv_weight, out_bias, out_scale, channels):
439
        args = [x, conv_weight, out_bias]
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458
        out_bias = relay.expand_dims(out_bias, axis=1, num_newaxis=2)
        y1 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y1 = relay.nn.relu(y1)
        y2 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1),
                             out_layout="CNHW")
        # fold will fail because the axis from two path
        # differs from each other.
        y2 = relay.nn.relu(y2)
        y = relay.add(y1, y2)
        y = relay.multiply(y, out_scale)
        return relay.Function(args, y)

    def fail2(x, conv_weight, out_bias, out_scale, channels):
459
        args = [x, conv_weight, out_bias]
460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
        out_bias = relay.expand_dims(out_bias, axis=1, num_newaxis=2)
        y1 = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             padding=(1, 1))
        y2 = relay.nn.relu(y1)
        # fold will fail because y1 is referred also by y2
        y1 = relay.multiply(y1, out_scale)
        y = relay.add(y1, y2)
        return relay.Function(args, y)

    def check(shape, channels, fbefore):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        weight = relay.var("weight")
        out_bias = relay.var("out_bias", shape=(channels,))
476
        out_scale = relay.const(_get_positive_scale((channels, 1, 1)))
477
        y1 = fbefore(x, weight, out_bias, out_scale, channels)
Zhi committed
478 479 480
        y1 = run_opt_pass(y1, transform.InferType())
        y1_folded = run_opt_pass(y1, transform.BackwardFoldScaleAxis())
        assert relay.analysis.alpha_equal(y1_folded, y1)
481 482 483 484 485

    check((4, 4, 10, 10), 4, fail1)
    check((4, 4, 10, 10), 4, fail2)


486 487 488 489 490 491 492 493 494 495
def test_fold_bwd_relu_fail():
    """testcase where we canont fold because scale can not pass relu"""
    def before(x, conv_weight, out_scale, channels):
        y = relay.nn.conv2d(x, conv_weight,
                             channels=channels,
                             kernel_size=(3, 3),
                             data_layout="NCHW",
                             padding=(1, 1))
        y = relay.nn.relu(y)
        y = relay.multiply(x, out_scale)
Zhi committed
496
        return relay.Function(relay.analysis.free_vars(y), y)
497 498 499 500 501 502

    def check(shape, channels, out_scale):
        x =  relay.var("x", shape=shape)
        in_channels = shape[1]
        weight = relay.var("weight")
        y1 = before(x, weight, out_scale, channels)
Zhi committed
503 504 505
        y1 = run_opt_pass(y1, transform.InferType())
        y1_folded = run_opt_pass(y1, transform.BackwardFoldScaleAxis())
        assert relay.analysis.alpha_equal(y1, y1_folded)
506 507 508 509 510 511 512

    out_scale = relay.var("in_scale", shape=(4, 1, 1))
    check((4, 4, 10, 10), 4, out_scale)
    out_scale = relay.const(np.random.uniform(size=(4, 1, 1), low=-1.0, high=0.0)).astype("float32")
    check((4, 4, 10, 10), 4, out_scale)


513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540
def test_fold_bwd_negative_scale():
    """Testcase of folding negative scale"""
    def before(x, conv_weight, out_scale, channels):
        args = [x, conv_weight]
        y = relay.nn.conv2d(x, conv_weight,
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        y = relay.multiply(y, out_scale)
        return relay.Function(args, y)

    def expected(x, conv_weight, out_scale, channels):
        # use a fixed order of args so alpha equal check can pass
        args = [x, conv_weight]
        squeezed_scale = relay.squeeze(out_scale, axis=[1,2])
        conv_weight = relay.multiply(
            conv_weight , relay.expand_dims(squeezed_scale, axis=1, num_newaxis=3))
        y = relay.nn.conv2d(x, conv_weight,
                            channels=channels,
                            kernel_size=(3, 3),
                            padding=(1, 1))
        return relay.Function(args, y)

    def check(shape, channels):
        x =  relay.var("x", shape=shape)
        weight = relay.var("weight")
        out_scale = relay.const(-_get_positive_scale((channels, 1, 1)))
        y1 = before(x, weight, out_scale, channels)
Zhi committed
541
        y1 = run_opt_pass(y1, transform.InferType())
542 543
        type_dict = {x.name_hint:x.checked_type for x in y1.params}
        weight = relay.var("weight", type_dict["weight"])
Zhi committed
544
        y1_folded = run_opt_pass(y1, transform.BackwardFoldScaleAxis())
545
        y1_expected = expected(x, weight, out_scale, channels)
Zhi committed
546 547
        y1_expected = run_opt_pass(y1_expected, transform.InferType())
        assert relay.analysis.alpha_equal(y1_folded, y1_expected)
548 549 550 551

    check((2, 4, 10, 10), 8)


552 553 554 555
if __name__ == "__main__":
    test_fold_fwd_simple()
    test_fold_fwd_dual_path()
    test_fold_fwd_fail()
556
    test_fold_fwd_relu_fail()
557
    test_fold_fwd_negative_scale()
558 559
    test_fold_bwd_simple()
    test_fold_bwd_dual_path()
560
    test_fold_bwd_dual_consumer()
561
    test_fold_bwd_fail()
562
    test_fold_bwd_relu_fail()
563
    test_fold_bwd_negative_scale()