onnx.cpp 53.5 KB
Newer Older
Paul's avatar
Paul committed
1
2
3
4
5
6
7
8
#include <google/protobuf/text_format.h>
#include <google/protobuf/io/zero_copy_stream_impl.h>
#include <onnx.pb.h>
#include <iostream>
#include <fstream>
#include <unordered_map>
#include <functional>
#include <array>
Paul's avatar
Paul committed
9
#include <utility>
10
#include <vector>
Paul's avatar
Paul committed
11

Paul's avatar
Paul committed
12
13
14
15
16
17
#include <migraphx/fallthrough.hpp>
#include <migraphx/program.hpp>
#include <migraphx/operators.hpp>
#include <migraphx/ranges.hpp>
#include <migraphx/instruction.hpp>
#include <migraphx/config.hpp>
18
#include <migraphx/onnx.hpp>
Paul's avatar
Paul committed
19
20

namespace migraphx {
Paul's avatar
Paul committed
21
inline namespace MIGRAPHX_INLINE_NS {
Paul's avatar
Paul committed
22
23
24
25
26

struct onnx_parser
{
    using attribute_map = std::unordered_map<std::string, onnx::AttributeProto>;
    using node_map      = std::unordered_map<std::string, onnx::NodeProto>;
Paul's avatar
Paul committed
27
28
    using op_func =
        std::function<std::vector<instruction_ref>(attribute_map, std::vector<instruction_ref>)>;
Paul's avatar
Paul committed
29
30
    node_map nodes;
    std::unordered_map<std::string, instruction_ref> instructions;
Scott Thornton's avatar
Scott Thornton committed
31
    program prog    = program();
32
    bool is_pytorch = false;
Paul's avatar
Paul committed
33
34

    std::unordered_map<std::string, op_func> ops;
35
    std::unordered_map<std::string, operation> map_actv_funcs;
Paul's avatar
Paul committed
36
37
38

    onnx_parser()
    {
Shucai Xiao's avatar
Shucai Xiao committed
39
        add_generic_op("MatMul", op::dot{});
Khalique's avatar
Khalique committed
40
        add_generic_op("Relu", op::relu{});
Khalique's avatar
Khalique committed
41
42
        add_generic_op("Sigmoid", op::sigmoid{});
        add_generic_op("Abs", op::abs{});
Shucai Xiao's avatar
Shucai Xiao committed
43
44
        add_generic_op("Exp", op::exp{});
        add_generic_op("Log", op::log{});
Khalique's avatar
Khalique committed
45
46
        // disable dropout for inference
        add_generic_op("Dropout", op::identity{});
Khalique's avatar
Khalique committed
47
        add_generic_op("Identity", op::identity{});
Shucai Xiao's avatar
Shucai Xiao committed
48
49
50
        add_generic_op("Sin", op::sin{});
        add_generic_op("Cos", op::cos{});
        add_generic_op("Tan", op::tan{});
51
52
        add_generic_op("Sinh", op::sinh{});
        add_generic_op("Cosh", op::cosh{});
53
        add_generic_op("Tanh", op::tanh{});
54
55
56
        add_generic_op("Asin", op::asin{});
        add_generic_op("Acos", op::acos{});
        add_generic_op("Atan", op::atan{});
Paul's avatar
Paul committed
57

Khalique's avatar
Khalique committed
58
59
60
61
62
        add_binary_op("Add", op::add{});
        add_binary_op("Div", op::div{});
        add_binary_op("Mul", op::mul{});
        add_binary_op("Sub", op::sub{});

Khalique's avatar
Khalique committed
63
64
65
        add_variadic_op("Sum", op::add{});
        add_variadic_op("Max", op::max{});
        add_variadic_op("Min", op::min{});
Paul's avatar
Paul committed
66

Khalique's avatar
Khalique committed
67
        add_mem_op("LRN", &onnx_parser::parse_lrn);
Khalique's avatar
Khalique committed
68
        add_mem_op("ImageScaler", &onnx_parser::parse_imagescaler);
69
        add_mem_op("LeakyRelu", &onnx_parser::parse_leaky_relu);
Khalique's avatar
Khalique committed
70
        add_mem_op("Elu", &onnx_parser::parse_elu);
Paul's avatar
Paul committed
71
72
        add_mem_op("Constant", &onnx_parser::parse_constant);
        add_mem_op("Conv", &onnx_parser::parse_conv);
Paul's avatar
Paul committed
73
74
        add_mem_op("MaxPool", &onnx_parser::parse_pooling);
        add_mem_op("AveragePool", &onnx_parser::parse_pooling);
75
76
        add_mem_op("GlobalMaxPool", &onnx_parser::parse_pooling);
        add_mem_op("GlobalAveragePool", &onnx_parser::parse_pooling);
Paul's avatar
Paul committed
77
        add_mem_op("Reshape", &onnx_parser::parse_reshape);
Paul's avatar
Paul committed
78
79
        add_mem_op("Flatten", &onnx_parser::parse_flatten);
        add_mem_op("Gemm", &onnx_parser::parse_gemm);
80
        add_mem_op("BatchNormalization", &onnx_parser::parse_batchnorm);
Paul's avatar
Paul committed
81
        add_mem_op("Softmax", &onnx_parser::parse_softmax);
Shucai Xiao's avatar
Shucai Xiao committed
82
        add_mem_op("LogSoftmax", &onnx_parser::parse_logsoftmax);
83
84
85
        add_mem_op("Squeeze", &onnx_parser::parse_squeeze);
        add_mem_op("Unsqueeze", &onnx_parser::parse_unsqueeze);
        add_mem_op("Slice", &onnx_parser::parse_slice);
Scott Thornton's avatar
Scott Thornton committed
86
        add_mem_op("Concat", &onnx_parser::parse_concat);
87
88
89
        add_mem_op("Gather", &onnx_parser::parse_gather);
        add_mem_op("Shape", &onnx_parser::parse_shape);
        add_mem_op("ConstantFill", &onnx_parser::parse_constant_fill);
Khalique's avatar
Khalique committed
90
        add_mem_op("Transpose", &onnx_parser::parse_transpose);
Shucai Xiao's avatar
Shucai Xiao committed
91
        add_mem_op("RNN", &onnx_parser::parse_rnn);
92
        add_mem_op("GRU", &onnx_parser::parse_gru);
Shucai Xiao's avatar
Shucai Xiao committed
93
        add_mem_op("LSTM", &onnx_parser::parse_lstm);
Khalique's avatar
Khalique committed
94
        add_mem_op("Pad", &onnx_parser::parse_pad);
95
96
97
98
99
100
101

        // init the activation function map
        init_actv_func();
    }

    void init_actv_func()
    {
102
103
104
105
106
        map_actv_funcs.insert(std::make_pair("tanh", op::tanh{}));
        map_actv_funcs.insert(std::make_pair("relu", op::relu{}));
        map_actv_funcs.insert(std::make_pair("sigmoid", op::sigmoid{}));
        map_actv_funcs.insert(std::make_pair("leakyrelu", op::leaky_relu{}));
        map_actv_funcs.insert(std::make_pair("elu", op::elu{}));
Paul's avatar
Paul committed
107
108
109
110
    }

    template <class F>
    void add_op(std::string name, F f)
Paul's avatar
Paul committed
111
112
113
114
115
116
117
118
119
    {
        ops.emplace(name, [=](auto&&... xs) {
            return std::vector<instruction_ref>{f(std::forward<decltype(xs)>(xs)...)};
        });
    }

    // Multi output op
    template <class F>
    void add_multi_op(std::string name, F f)
Paul's avatar
Paul committed
120
121
122
123
124
125
126
    {
        ops.emplace(name, f);
    }

    template <class F>
    void add_mem_op(std::string name, F f)
    {
Paul's avatar
Paul committed
127
        add_op(name, [=](auto&&... xs) {
Paul's avatar
Paul committed
128
129
130
            return std::mem_fn(f)(*this, name, std::forward<decltype(xs)>(xs)...);
        });
    }
Khalique's avatar
Khalique committed
131

132
    template <class T>
Khalique's avatar
Khalique committed
133
    void add_binary_op(std::string name, T x)
134
    {
Paul's avatar
Paul committed
135
        add_op(name, [this, x](attribute_map attributes, std::vector<instruction_ref> args) {
Scott Thornton's avatar
Scott Thornton committed
136
            if(args.size() != 2)
Paul's avatar
Paul committed
137
                MIGRAPHX_THROW("binary operators should have 2 operands");
138
            if(contains(attributes, "broadcast") and contains(attributes, "axis"))
139
140
141
142
            {
                uint64_t broadcasted = parse_value(attributes.at("broadcast")).at<uint64_t>();
                if(broadcasted != 0)
                {
143
                    uint64_t axis = parse_value(attributes.at("axis")).at<uint64_t>();
144
145
146
147
                    auto l =
                        prog.add_instruction(op::broadcast{axis, args[0]->get_shape()}, args[1]);
                    return prog.add_instruction(x, args[0], l);
                }
148
                return prog.add_instruction(x, args);
149
            }
Paul's avatar
Paul committed
150
            else
151
            {
Khalique's avatar
Khalique committed
152
                return add_broadcastable_binary_op(args[0], args[1], x);
153
154
155
156
            }
        });
    }

Khalique's avatar
Khalique committed
157
158
159
160
161
    template <class T>
    instruction_ref add_broadcastable_binary_op(instruction_ref arg0, instruction_ref arg1, T x)
    {
        if(arg0->get_shape() != arg1->get_shape())
        {
Khalique's avatar
Khalique committed
162
163
164
165
166
167
168
169
170
171
172
173
174
            // Example:
            // s0 = (3,2,4,5) and s1 = (2,1,1)
            //
            // In this case we need to broadcast (:,1,1) portion of
            // s1 plus broadcast the 1st dimension of s1
            // giving output_lens = (3,2,4,5)
            //
            // Another example:
            // s0 = (3,2,1,5) and s1 = (2,7,5)
            // In this case we need to broadcast the (:,:,1:,:) axis
            // of s0 plus the 1st dimension of s1 giving
            // output_lens = (3,2,7,5)
            //
Khalique's avatar
Khalique committed
175
176
177
178
179
180
181
182
            // Get lengths for both arguments
            const std::vector<std::size_t>* s0 = &arg0->get_shape().lens();
            const std::vector<std::size_t>* s1 = &arg1->get_shape().lens();

            // Make sure s0 is the smaller size
            if(s0->size() > s1->size())
                std::swap(s0, s1);

Khalique's avatar
Khalique committed
183
            std::vector<std::size_t> output_lens(*s1);
Khalique's avatar
Khalique committed
184
185
            auto offset = s1->size() - s0->size();
            std::transform(s0->begin(),
Khalique's avatar
Khalique committed
186
187
188
189
                           s0->end(),
                           s1->begin() + offset,
                           output_lens.begin() + offset,
                           [](auto a, auto b) { return std::max(a, b); });
Khalique's avatar
Khalique committed
190
191
192
193
194
195
196
197
198

            auto l0 = prog.add_instruction(op::multibroadcast{output_lens}, arg0);
            auto l1 = prog.add_instruction(op::multibroadcast{output_lens}, arg1);
            return prog.add_instruction(x, l0, l1);
        }
        else
        {
            return prog.add_instruction(x, {arg0, arg1});
        }
199
200
    }

Paul's avatar
Paul committed
201
    template <class T>
Paul's avatar
Paul committed
202
203
    void add_generic_op(std::string name, T x)
    {
Paul's avatar
Paul committed
204
        add_op(name, [this, x](attribute_map, std::vector<instruction_ref> args) {
Paul's avatar
Paul committed
205
206
207
208
            return prog.add_instruction(x, args);
        });
    }

Khalique's avatar
Khalique committed
209
    template <class T>
Khalique's avatar
Khalique committed
210
    void add_variadic_op(std::string name, T x)
Khalique's avatar
Khalique committed
211
    {
Paul's avatar
Paul committed
212
        add_op(name, [this, x](attribute_map, std::vector<instruction_ref> args) {
Khalique's avatar
Khalique committed
213
            return std::accumulate(std::next(args.begin()),
Khalique's avatar
Khalique committed
214
215
216
217
218
                                   args.end(),
                                   args.front(),
                                   [this, x](instruction_ref a, instruction_ref b) {
                                       return add_broadcastable_binary_op(a, b, x);
                                   });
Khalique's avatar
Khalique committed
219
        });
Khalique's avatar
Khalique committed
220
221
    }

Paul's avatar
Paul committed
222
    instruction_ref
Paul's avatar
Paul committed
223
    parse_softmax(const std::string&, const attribute_map&, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
224
225
    {
        auto dims = args.front()->get_shape().lens();
Scott Thornton's avatar
Scott Thornton committed
226
227
        auto r =
            prog.add_instruction(op::reshape{{long(dims[0]), long(dims[1]), 1, 1}}, args.front());
228
229
        auto s = prog.add_instruction(op::softmax{}, r);
        return prog.add_instruction(op::reshape{{long(dims[0]), long(dims[1])}}, s);
Paul's avatar
Paul committed
230
231
    }

Shucai Xiao's avatar
Shucai Xiao committed
232
233
234
    instruction_ref parse_logsoftmax(const std::string&,
                                     const attribute_map& attributes,
                                     std::vector<instruction_ref> args)
Shucai Xiao's avatar
Shucai Xiao committed
235
236
237
238
239
240
241
242
243
244
    {
        int axis = 1;
        if(contains(attributes, "axis"))
        {
            axis = parse_value(attributes.at("axis")).at<int>();
        }

        return prog.add_instruction(op::logsoftmax{axis}, std::move(args));
    }

Paul's avatar
Paul committed
245
    instruction_ref
Paul's avatar
Paul committed
246
    parse_conv(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
247
    {
248
        op::convolution op;
249
        auto l0 = args[0];
Paul's avatar
Paul committed
250
251
        if(contains(attributes, "pads"))
        {
Scott Thornton's avatar
Scott Thornton committed
252
            if(contains(attributes, "auto_pad"))
253
            {
Paul's avatar
Paul committed
254
                MIGRAPHX_THROW("auto_pad and padding cannot be specified simultaneously");
255
            }
256
257
            std::vector<std::int64_t> padding;
            copy(attributes["pads"].ints(), std::back_inserter(padding));
Scott Thornton's avatar
Scott Thornton committed
258
            if(padding.size() != 4)
259
            {
Paul's avatar
Paul committed
260
                MIGRAPHX_THROW("padding should have 4 values");
261
            }
Scott Thornton's avatar
Scott Thornton committed
262
            if(padding[0] != padding[2] || padding[1] != padding[3])
263
            {
264
265
                // insert zeros for pad op (args[0] has 4 dims)
                padding = {0, 0, padding[0], padding[1], 0, 0, padding[2], padding[3]};
Khalique's avatar
Khalique committed
266
                l0      = prog.add_instruction(op::pad{padding}, l0);
267
            }
268
269
270
271
            else
            {
                op.padding[0] = padding[0];
                op.padding[1] = padding[1];
272
            }
Paul's avatar
Paul committed
273
        }
Paul's avatar
Paul committed
274
275
276
277
278
279
280
281
        if(contains(attributes, "strides"))
        {
            copy(attributes["strides"].ints(), op.stride.begin());
        }
        if(contains(attributes, "dilations"))
        {
            copy(attributes["dilations"].ints(), op.dilation.begin());
        }
Scott Thornton's avatar
Scott Thornton committed
282
        if(contains(attributes, "auto_pad"))
283
284
        {
            auto s = attributes["auto_pad"].s();
Scott Thornton's avatar
Scott Thornton committed
285
            if(contains(attributes, "pads") and to_upper(s) != "NOTSET")
286
            {
Paul's avatar
Paul committed
287
                MIGRAPHX_THROW("auto_pad and padding cannot be specified simultaneously");
288
289
            }

wsttiger's avatar
fixes  
wsttiger committed
290
            if(s.find("SAME") != std::string::npos)
291
            {
292
                op.padding_mode = op::padding_mode_t::same;
293
294
            }
        }
Khalique's avatar
Khalique committed
295
296
297
298
        if(contains(attributes, "group"))
        {
            op.group = parse_value(attributes.at("group")).at<int>();
        }
Paul's avatar
Paul committed
299
300
301
302
        if(args.size() == 3)
        {
            uint64_t axis = 1;
            auto l1       = prog.add_instruction(op, args[0], args[1]);
Scott Thornton's avatar
Scott Thornton committed
303
            auto l2       = prog.add_instruction(op::broadcast{axis, l1->get_shape()}, args[2]);
304
            return prog.add_instruction(op::add{}, l1, l2);
Paul's avatar
Paul committed
305
        }
306
        return prog.add_instruction(op, l0, args[1]);
Paul's avatar
Paul committed
307
    }
Paul's avatar
Paul committed
308

Paul's avatar
Paul committed
309
310
311
    instruction_ref parse_pooling(const std::string& name,
                                  attribute_map attributes,
                                  std::vector<instruction_ref> args)
Paul's avatar
Paul committed
312
    {
Khalique's avatar
Khalique committed
313
        op::pooling op{ends_with(name, "MaxPool") ? "max" : "average"};
314
        auto l0 = args[0];
Khalique's avatar
Khalique committed
315
        if(starts_with(name, "Global"))
316
        {
Khalique's avatar
Khalique committed
317
318
            auto lens  = args.front()->get_shape().lens();
            op.lengths = {lens[2], lens[3]};
319
        }
Paul's avatar
Paul committed
320
321
        if(contains(attributes, "pads"))
        {
322
323
            std::vector<std::int64_t> padding;
            copy(attributes["pads"].ints(), std::back_inserter(padding));
Scott Thornton's avatar
Scott Thornton committed
324
            if(padding.size() != 4)
325
            {
Paul's avatar
Paul committed
326
                MIGRAPHX_THROW("padding should have 4 values");
327
            }
Scott Thornton's avatar
Scott Thornton committed
328
            if(padding[0] != padding[2] || padding[1] != padding[3])
329
            {
330
331
                // insert zeros for pad op (args[0] has 4 dims)
                padding = {0, 0, padding[0], padding[1], 0, 0, padding[2], padding[3]};
Khalique's avatar
Khalique committed
332
                l0      = prog.add_instruction(op::pad{padding}, l0);
333
334
335
336
337
            }
            else
            {
                op.padding[0] = padding[0];
                op.padding[1] = padding[1];
338
            }
Paul's avatar
Paul committed
339
340
341
342
343
344
345
346
347
        }
        if(contains(attributes, "strides"))
        {
            copy(attributes["strides"].ints(), op.stride.begin());
        }
        if(contains(attributes, "kernel_shape"))
        {
            copy(attributes["kernel_shape"].ints(), op.lengths.begin());
        }
Scott Thornton's avatar
Scott Thornton committed
348
        if(contains(attributes, "auto_pad"))
349
350
        {
            auto s = attributes["auto_pad"].s();
351
            if(s.find("SAME_UPPER") == std::string::npos)
352
            {
353
                MIGRAPHX_THROW("auto_pad only supports SAME_UPPER for pooling");
354
            }
355
            op.padding_mode = op::padding_mode_t::same;
356
357
        }

358
        return prog.add_instruction(op, l0);
Paul's avatar
Paul committed
359
360
    }

Paul's avatar
Paul committed
361
    instruction_ref
Paul's avatar
Paul committed
362
    parse_reshape(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
363
    {
364
        op::reshape op;
Paul's avatar
Paul committed
365
366
367
368
369
370
371
        if(args.size() == 1)
        {
            literal s = parse_value(attributes.at("shape"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.dims)); });
        }
        if(args.size() == 2)
        {
Paul's avatar
Paul committed
372
            auto s = args[1]->eval();
Paul's avatar
Paul committed
373
            if(s.empty())
Paul's avatar
Paul committed
374
                MIGRAPHX_THROW("Dynamic shape is not supported.");
Paul's avatar
Paul committed
375
            s.visit([&](auto v) { copy(v, std::back_inserter(op.dims)); });
Paul's avatar
Paul committed
376
        }
Paul's avatar
Paul committed
377
378
379
        return prog.add_instruction(op, args[0]);
    }

Paul's avatar
Paul committed
380
    instruction_ref
Paul's avatar
Paul committed
381
    parse_flatten(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
382
    {
383
        uint64_t axis = 1;
Paul's avatar
Paul committed
384
385
386
387
        if(contains(attributes, "axis"))
        {
            axis = parse_value(attributes.at("axis")).at<int>();
        }
388
        return prog.add_instruction(op::flatten{axis}, args[0]);
Paul's avatar
Paul committed
389
390
    }

391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
    instruction_ref
    parse_squeeze(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        op::squeeze op;
        literal s = parse_value(attributes.at("axes"));
        s.visit([&](auto v) { copy(v, std::back_inserter(op.axes)); });
        return prog.add_instruction(op, args[0]);
    }

    instruction_ref
    parse_unsqueeze(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        op::unsqueeze op;
        literal s = parse_value(attributes.at("axes"));
        s.visit([&](auto v) { copy(v, std::back_inserter(op.axes)); });
        return prog.add_instruction(op, args[0]);
    }

Scott Thornton's avatar
Scott Thornton committed
409
410
411
412
413
414
415
    instruction_ref
    parse_concat(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        std::size_t axis = parse_value(attributes.at("axis")).at<int>();
        op::concat op{axis};
        return prog.add_instruction(op, std::move(args));
    }
416

417
418
419
    instruction_ref
    parse_gather(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
420
        int axis = 0;
421
422
423
424
        if(contains(attributes, "axis"))
        {
            axis = parse_value(attributes.at("axis")).at<int>();
        }
425
        op::gather op{axis};
426
427
428
        return prog.add_instruction(op, std::move(args));
    }

429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
    instruction_ref
    parse_slice(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        op::slice op;
        if(contains(attributes, "axes"))
        {
            literal s = parse_value(attributes.at("axes"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.axes)); });
        }
        {
            literal s = parse_value(attributes.at("ends"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.ends)); });
        }
        {
            literal s = parse_value(attributes.at("starts"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.starts)); });
        }
        return prog.add_instruction(op, args[0]);
    }

Paul's avatar
Paul committed
449
450
451
    instruction_ref parse_constant(const std::string&,
                                   attribute_map attributes,
                                   const std::vector<instruction_ref>&)
Paul's avatar
Paul committed
452
    {
Shucai Xiao's avatar
Shucai Xiao committed
453
        literal v     = parse_value(attributes.at("value"));
454
455
456
        auto dim_size = attributes.at("value").t().dims_size();
        // if dim_size is 0, it is a scalar
        if(dim_size == 0)
457
        {
458
            migraphx::shape scalar_shape{v.get_shape().type()};
459
460
461
            return prog.add_literal(migraphx::literal{scalar_shape, v.data()});
        }

Paul's avatar
Paul committed
462
463
        return prog.add_literal(v);
    }
Paul's avatar
Paul committed
464

Paul's avatar
Paul committed
465
    instruction_ref
Paul's avatar
Paul committed
466
    parse_gemm(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
467
468
    {
        float alpha = 1.0f;
Khalique's avatar
Khalique committed
469
        float beta  = 1.0f;
Paul's avatar
Paul committed
470
471
472
473
474
475
476
477
        bool transa = false;
        bool transb = false;
        if(contains(attributes, "alpha"))
        {
            alpha = parse_value(attributes.at("alpha")).at<float>();
        }
        if(contains(attributes, "beta"))
        {
478
            beta = parse_value(attributes.at("beta")).at<float>();
Paul's avatar
Paul committed
479
480
481
482
483
484
485
486
487
        }
        if(contains(attributes, "transA"))
        {
            transa = parse_value(attributes.at("transA")).at<bool>();
        }
        if(contains(attributes, "transB"))
        {
            transb = parse_value(attributes.at("transB")).at<bool>();
        }
488
489
490
491
492
493

        std::vector<int64_t> perm(args[0]->get_shape().lens().size());
        std::iota(perm.begin(), perm.end(), int64_t{0});
        // swap the last two elements
        std::swap(*perm.rbegin(), *(perm.rbegin() + 1));

494
495
        auto l1 = (transa) ? prog.add_instruction(op::transpose{perm}, args[0]) : args[0];
        auto l2 = (transb) ? prog.add_instruction(op::transpose{perm}, args[1]) : args[1];
Paul's avatar
Paul committed
496
497
        if(args.size() == 3)
        {
498
            if(beta != 0.f && args[2]->get_shape().elements() > 0)
499
            {
Shucai Xiao's avatar
Shucai Xiao committed
500
                auto out_lens   = l1->get_shape().lens();
501
                out_lens.back() = l2->get_shape().lens().back();
Shucai Xiao's avatar
Shucai Xiao committed
502
503
504
                auto l3         = args[2];
                if(!std::equal(
                       out_lens.begin(), out_lens.end(), args[2]->get_shape().lens().begin()))
Khalique's avatar
Khalique committed
505
                {
506
                    l3 = prog.add_instruction(op::multibroadcast{out_lens}, args[2]);
Khalique's avatar
Khalique committed
507
                }
508
509

                return prog.add_instruction(op::dot{alpha, beta}, l1, l2, l3);
510
            }
Paul's avatar
Paul committed
511
        }
512
        return prog.add_instruction(op::dot{alpha}, l1, l2);
Paul's avatar
Paul committed
513
514
    }

515
    instruction_ref
Paul's avatar
Paul committed
516
    parse_batchnorm(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
517
    {
Scott Thornton's avatar
Scott Thornton committed
518
519
        float epsilon                                     = 1e-5f;
        float momentum                                    = 0.9f;
520
        op::batch_norm_inference::bn_infer_mode_t bn_mode = op::batch_norm_inference::spatial;
Scott Thornton's avatar
Scott Thornton committed
521
        bool is_test                                      = false;
522
523
524
525
526
527
        if(contains(attributes, "epsilon"))
        {
            epsilon = parse_value(attributes.at("epsilon")).at<float>();
        }
        if(contains(attributes, "momentum"))
        {
528
            momentum = parse_value(attributes.at("momentum")).at<float>();
529
530
531
        }
        if(contains(attributes, "is_test"))
        {
wsttiger's avatar
wsttiger committed
532
            is_test = parse_value(attributes.at("is_test")).at<uint64_t>() > 0;
533
534
535
        }
        if(contains(attributes, "spatial"))
        {
536
            bn_mode = (parse_value(attributes.at("spatial")).at<uint64_t>() > 0)
537
538
                          ? op::batch_norm_inference::spatial
                          : op::batch_norm_inference::per_activation;
539
        }
Paul's avatar
Paul committed
540
        (void)is_test;
Paul's avatar
Paul committed
541
        op::batch_norm_inference op{epsilon, momentum, bn_mode};
Paul's avatar
Paul committed
542
        return prog.add_instruction(op, std::move(args));
543
544
    }

545
546
547
548
    instruction_ref parse_leaky_relu(const std::string&,
                                     attribute_map attributes,
                                     std::vector<instruction_ref> args)
    {
Khalique's avatar
Khalique committed
549
        float alpha = 0.01; // default alpha val for leaky relu
550
551
552
553
554
555
556
557
        if(contains(attributes, "alpha"))
        {
            alpha = parse_value(attributes.at("alpha")).at<float>();
        }
        op::leaky_relu op{alpha};
        return prog.add_instruction(op, args.front());
    }

Khalique's avatar
Khalique committed
558
559
    instruction_ref
    parse_elu(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Khalique's avatar
Khalique committed
560
561
562
563
564
565
566
567
568
569
    {
        float alpha = 1.0; // default alpha val for elu
        if(contains(attributes, "alpha"))
        {
            alpha = parse_value(attributes.at("alpha")).at<float>();
        }
        op::elu op{alpha};
        return prog.add_instruction(op, args.front());
    }

Khalique's avatar
Khalique committed
570
571
    instruction_ref
    parse_lrn(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Khalique's avatar
Khalique committed
572
573
    {
        float alpha = 0.0001;
Khalique's avatar
Khalique committed
574
575
576
        float beta  = 0.75;
        float bias  = 1.0;
        int size    = 1;
Khalique's avatar
Khalique committed
577
578
579
580
581
582
583
584
585
586
587
588
        if(contains(attributes, "alpha"))
            alpha = parse_value(attributes.at("alpha")).at<float>();
        if(contains(attributes, "beta"))
            beta = parse_value(attributes.at("beta")).at<float>();
        if(contains(attributes, "bias"))
            bias = parse_value(attributes.at("bias")).at<float>();
        if(contains(attributes, "size"))
            size = parse_value(attributes.at("size")).at<int>();
        op::lrn op{alpha, beta, bias, size};
        return prog.add_instruction(op, args.front());
    }

Khalique's avatar
Khalique committed
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
    instruction_ref parse_imagescaler(const std::string&,
                                      attribute_map attributes,
                                      std::vector<instruction_ref> args)
    {
        float scale = 1.0;
        std::vector<float> bias{};
        if(contains(attributes, "scale"))
        {
            scale = parse_value(attributes.at("scale")).at<float>();
        }

        if(contains(attributes, "bias"))
        {
            auto&& bias_floats = attributes["bias"].floats();
            bias               = std::vector<float>(bias_floats.begin(), bias_floats.end());
        }
        auto input_shape = args.front()->get_shape();
Khalique's avatar
Khalique committed
606

Khalique's avatar
Khalique committed
607
608
        auto scale_val = prog.add_literal(scale);
        auto bias_vals = prog.add_literal(
Paul's avatar
Paul committed
609
            migraphx::literal{migraphx::shape{migraphx::shape::float_type, {bias.size()}}, bias});
Khalique's avatar
Khalique committed
610

Paul's avatar
Paul committed
611
612
        auto scale_tensor = prog.add_instruction(migraphx::op::scalar{input_shape}, scale_val);
        auto img_scaled   = prog.add_instruction(migraphx::op::mul{}, args.front(), scale_tensor);
Paul's avatar
Paul committed
613
        auto bias_bcast = prog.add_instruction(migraphx::op::broadcast{1, input_shape}, bias_vals);
Paul's avatar
Paul committed
614
        return prog.add_instruction(migraphx::op::add{}, img_scaled, bias_bcast);
Khalique's avatar
Khalique committed
615
    }
Khalique's avatar
Khalique committed
616

Khalique's avatar
Khalique committed
617
618
    instruction_ref
    parse_transpose(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Khalique's avatar
Khalique committed
619
620
621
622
623
624
625
    {
        std::vector<int64_t> perm{};
        if(contains(attributes, "perm"))
        {
            auto&& perm_vals = attributes["perm"].ints();
            perm             = std::vector<int64_t>(perm_vals.begin(), perm_vals.end());
        }
Paul's avatar
Paul committed
626
        return prog.add_instruction(migraphx::op::transpose{perm}, args.front());
Khalique's avatar
Khalique committed
627
628
    }

Khalique's avatar
Khalique committed
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
    instruction_ref
    parse_pad(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        std::vector<int64_t> pads{};
        float value = 0.0f;
        if(contains(attributes, "pads"))
        {
            auto&& pad_vals = attributes["pads"].ints();
            pads            = std::vector<int64_t>(pad_vals.begin(), pad_vals.end());
        }
        if(contains(attributes, "value"))
        {
            value = parse_value(attributes.at("value")).at<float>();
        }
        if(contains(attributes, "mode"))
        {
            auto mode = attributes.at("mode").s();
            if(mode != "constant")
                MIGRAPHX_THROW("migraphx currently only supports constant padding");
        }
        return prog.add_instruction(migraphx::op::pad{pads, value}, args.front());
    }
651
652
653
    // Use a literal instruction to replace the shape since, output of
    // shape operator are literals in migraphx
    instruction_ref
Shucai Xiao's avatar
Shucai Xiao committed
654
    parse_shape(const std::string&, const attribute_map&, std::vector<instruction_ref> args)
655
656
    {
        if(args.size() != 1)
657
            MIGRAPHX_THROW("Shape: operator should have 1 operand");
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
        std::vector<std::size_t> arg_shape = args[0]->get_shape().lens();
        std::vector<int64_t> vec_shape(arg_shape.size());
        migraphx::shape s(migraphx::shape::int64_type, {arg_shape.size()});
        std::transform(arg_shape.begin(), arg_shape.end(), vec_shape.begin(), [](auto i) {
            return int64_t(i);
        });
        return prog.add_literal(migraphx::literal{s, vec_shape});
    }

    // Use a literal instruction to replace the constantFill operator. In RNN, input shape
    // and value are fixed, so no need to do the actual computation for the constantFill
    // operator
    instruction_ref parse_constant_fill(const std::string&,
                                        attribute_map attributes,
                                        std::vector<instruction_ref> args)
    {
        int input_as_shape = 0;
        int dtype          = 1;
        float value        = 0.0f;

        if(contains(attributes, "dtype"))
        {
            dtype = parse_value(attributes.at("dtype")).at<int>();
        }
        migraphx::shape::type_t type = get_type(dtype);

        if(contains(attributes, "input_as_shape"))
        {
            input_as_shape = parse_value(attributes.at("input_as_shape")).at<int>();
        }

        if(contains(attributes, "value"))
        {
            value = parse_value(attributes.at("value")).at<float>();
        }

Shucai Xiao's avatar
Shucai Xiao committed
694
695
        if(contains(attributes, "extra_shape"))
        {
696
            MIGRAPHX_THROW("ConstantFill: cannot handle extra shape attribute");
697
698
        }

699
700
        if(input_as_shape == 1)
        {
Shucai Xiao's avatar
Shucai Xiao committed
701
            if(args.size() != 1)
702
            {
703
                MIGRAPHX_THROW("ConstantFill: need an input argument as output shape");
704
705
            }

Shucai Xiao's avatar
Shucai Xiao committed
706
707
            if(contains(attributes, "shape"))
            {
708
                MIGRAPHX_THROW("ConstantFill: cannot set the shape argument and pass in an input "
Shucai Xiao's avatar
Shucai Xiao committed
709
                               "at the same time");
710
711
            }

712
713
714
            migraphx::argument in = args[0]->eval();
            if(in.empty())
            {
715
                MIGRAPHX_THROW("ConstantFill: cannot handle dynamic shape as input");
716
            }
717

718
719
720
            std::vector<std::size_t> dims;
            in.visit([&](auto input) { dims.assign(input.begin(), input.end()); });
            migraphx::shape s(type, dims);
721
722
            std::vector<float> values(s.elements(), value);
            return prog.add_literal(migraphx::literal(s, values));
723
724
725
        }
        else if(input_as_shape == 0)
        {
Shucai Xiao's avatar
Shucai Xiao committed
726
727
            if(!contains(attributes, "shape"))
            {
728
                MIGRAPHX_THROW("ConstantFill: attribute output shape is needed");
729
730
731
            }

            literal ls = parse_value(attributes.at("shape"));
732
            std::vector<std::size_t> dims;
Shucai Xiao's avatar
Shucai Xiao committed
733
            ls.visit([&](auto s) { dims.assign(s.begin(), s.end()); });
734
            migraphx::shape s{type, dims};
735
736
            std::vector<float> values(s.elements(), value);
            return prog.add_literal(migraphx::literal(s, values));
737
738
739
        }
        else
        {
740
            MIGRAPHX_THROW("ConstantFill: wrong value of attribute input_as_shape");
741
742
743
        }
    }

Shucai Xiao's avatar
Shucai Xiao committed
744
    std::vector<instruction_ref>
Shucai Xiao's avatar
Shucai Xiao committed
745
746
747
    parse_rnn(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        migraphx::shape input_shape = args[0]->get_shape();
748
        std::size_t hidden_size     = args[1]->get_shape().lens()[1];
Shucai Xiao's avatar
Shucai Xiao committed
749
750
751

        if(contains(attributes, "hidden_size"))
        {
Shucai Xiao's avatar
Shucai Xiao committed
752
            std::size_t hidden_size_att = parse_value(attributes.at("hidden_size")).at<int>();
Shucai Xiao's avatar
Shucai Xiao committed
753
            if(hidden_size != hidden_size_att)
Shucai Xiao's avatar
Shucai Xiao committed
754
755
756
            {
                MIGRAPHX_THROW("RNN: hidden size mismatch in input and attribute");
            }
Shucai Xiao's avatar
Shucai Xiao committed
757
758
759
760
761
762
763
764
765
        }

        // Handling of direction to be added later
        std::string direction{"forward"};
        if(contains(attributes, "direction"))
        {
            direction = attributes.at("direction").s();
        }

766
        op::rnn_direction dirct = op::rnn_direction::forward;
Shucai Xiao's avatar
Shucai Xiao committed
767
768
        if(direction == "bidirectional")
        {
769
            dirct = op::rnn_direction::bidirectional;
Shucai Xiao's avatar
Shucai Xiao committed
770
771
772
        }
        else if(direction == "reverse")
        {
773
            dirct = op::rnn_direction::reverse;
Shucai Xiao's avatar
Shucai Xiao committed
774
775
        }

776
777
778
779
780
        std::vector<std::string> vec_names{"tanh"};
        if(contains(attributes, "activations"))
        {
            auto names = attributes.at("activations").strings();
            vec_names.clear();
781
            vec_names.resize(names.size());
782
            std::copy(names.begin(), names.end(), vec_names.begin());
783
784
        }

785
786
787
        auto name_it = std::find_if(vec_names.begin(), vec_names.end(), [&](auto& name) {
            return (map_actv_funcs.count(name) == 0);
        });
Shucai Xiao's avatar
Shucai Xiao committed
788
        if(name_it != vec_names.end())
789
790
791
        {
            MIGRAPHX_THROW("RNN: activation function " + std::string(*name_it) + " not supported");
        }
792

Shucai Xiao's avatar
Shucai Xiao committed
793
        // bidirectional case should have two activation functions.
Shucai Xiao's avatar
Shucai Xiao committed
794
        // one is for forward, and the other is for reverse.
Shucai Xiao's avatar
Shucai Xiao committed
795
        // if only one actv function is provided, we use it in both
796
        // forward and reverse direction
797
        if(dirct == op::rnn_direction::bidirectional)
798
        {
Shucai Xiao's avatar
Shucai Xiao committed
799
            if(vec_names.size() == 1)
800
801
802
803
804
            {
                vec_names.push_back(vec_names.at(0));
            }
        }

Shucai Xiao's avatar
Shucai Xiao committed
805
806
807
        std::vector<operation> vec_actv_funcs(vec_names.size());
        std::transform(vec_names.begin(), vec_names.end(), vec_actv_funcs.begin(), [&](auto& fn) {
            return map_actv_funcs[fn];
808
        });
Shucai Xiao's avatar
Shucai Xiao committed
809

Shucai Xiao's avatar
Shucai Xiao committed
810
811
812
813
814
815
816
        // To be added later
        float clip = 0.0;
        if(contains(attributes, "clip"))
        {
            clip = parse_value(attributes.at("clip")).at<float>();
        }

817
818
        // if the number of arguments is less than 6, append
        // undefined operator to have 6 arguments
Shucai Xiao's avatar
Shucai Xiao committed
819
        if(args.size() < 6)
820
821
822
823
824
        {
            auto ins = prog.add_instruction(op::undefined{});
            args.insert(args.end(), (6 - args.size()), ins);
        }

Shucai Xiao's avatar
Shucai Xiao committed
825
826
        // first output for the concatenation of hidden states
        auto hidden_states = prog.add_instruction(op::rnn{hidden_size, vec_actv_funcs, dirct, clip},
Shucai Xiao's avatar
Shucai Xiao committed
827
                                                  std::move(args));
Shucai Xiao's avatar
Shucai Xiao committed
828

829
        // second output for the last hidden state
Shucai Xiao's avatar
Shucai Xiao committed
830
        auto last_output = prog.add_instruction(op::rnn_last_output{}, hidden_states);
Shucai Xiao's avatar
Shucai Xiao committed
831

Shucai Xiao's avatar
Shucai Xiao committed
832
        return {hidden_states, last_output};
Shucai Xiao's avatar
Shucai Xiao committed
833
834
    }

835
    std::vector<instruction_ref>
836
837
838
839
840
841
842
    parse_gru(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        migraphx::shape input_shape = args[0]->get_shape();
        std::size_t hidden_size     = args[2]->get_shape().lens()[2];

        if(contains(attributes, "hidden_size"))
        {
Shucai Xiao's avatar
Shucai Xiao committed
843
            std::size_t hidden_size_att = parse_value(attributes.at("hidden_size")).at<int>();
Shucai Xiao's avatar
Shucai Xiao committed
844
            if(hidden_size != hidden_size_att)
Shucai Xiao's avatar
Shucai Xiao committed
845
846
847
            {
                MIGRAPHX_THROW("GRU: hidden size mismatch in input and attribute");
            }
848
849
850
851
852
853
854
855
856
        }

        // Handling of direction to be added later
        std::string direction{"forward"};
        if(contains(attributes, "direction"))
        {
            direction = attributes.at("direction").s();
        }

857
        op::rnn_direction dirct = op::rnn_direction::forward;
858
859
        if(direction == "bidirectional")
        {
860
            dirct = op::rnn_direction::bidirectional;
861
862
863
        }
        else if(direction == "reverse")
        {
864
            dirct = op::rnn_direction::reverse;
865
866
        }

867
        std::vector<std::string> vec_names = {"sigmoid", "tanh"};
868
869
        if(contains(attributes, "activations"))
        {
870
            auto names = attributes.at("activations").strings();
871
            vec_names.clear();
Shucai Xiao's avatar
Shucai Xiao committed
872
            vec_names.resize(names.size());
873
            std::copy(names.begin(), names.end(), vec_names.begin());
874
875
        }

876
        // need 4 activation functions
877
        if(dirct == op::rnn_direction::bidirectional)
878
        {
Shucai Xiao's avatar
Shucai Xiao committed
879
            // 4 activation functions are used in the bidirectional
880
            // scenario. No spec is provided in onnx::operator. we
Shucai Xiao's avatar
Shucai Xiao committed
881
882
            // use the algorithm that: if 1 actv function is provided,
            // repeat 1 four times. If 2 actv functins are provided,
883
884
            // assume forward and reverse use the same pair of actv
            // functions. For the case of 3 actv functions provided,
Shucai Xiao's avatar
Shucai Xiao committed
885
886
887
            // assume the 3rd one is repeated once and used by the
            // reverse direction.
            // This may need change later
888
            if(vec_names.size() == 1)
889
            {
890
                vec_names.insert(vec_names.end(), 3, vec_names.at(0));
891
            }
892
            else if(vec_names.size() == 2)
893
            {
894
895
896
                // repeat the activation functions
                vec_names.push_back(vec_names.at(0));
                vec_names.push_back(vec_names.at(1));
897
            }
898
            else if(vec_names.size() == 3)
899
            {
900
                vec_names.push_back(vec_names.at(2));
901
902
            }
        }
Shucai Xiao's avatar
Shucai Xiao committed
903
        else
904
        {
905
            if(vec_names.size() == 1)
906
            {
907
                vec_names.push_back(vec_names.at(0));
908
909
910
            }
        }

911
912
913
        auto name_it = std::find_if(vec_names.begin(), vec_names.end(), [&](auto& name) {
            return (map_actv_funcs.count(name) == 0);
        });
Shucai Xiao's avatar
Shucai Xiao committed
914
        if(name_it != vec_names.end())
915
916
917
        {
            MIGRAPHX_THROW("GRU: activation function " + std::string(*name_it) + " not supported");
        }
918

Shucai Xiao's avatar
Shucai Xiao committed
919
920
921
        std::vector<operation> vec_actv_funcs(vec_names.size());
        std::transform(vec_names.begin(), vec_names.end(), vec_actv_funcs.begin(), [&](auto& name) {
            return map_actv_funcs[name];
Shucai Xiao's avatar
Shucai Xiao committed
922
        });
923
924
925
926
927
928
929
930

        float clip = 0.0;
        if(contains(attributes, "clip"))
        {
            clip = parse_value(attributes.at("clip")).at<float>();
        }

        int linear_before_reset = 0;
Shucai Xiao's avatar
Shucai Xiao committed
931
        if(contains(attributes, "linear_before_reset"))
932
933
934
935
        {
            linear_before_reset = parse_value(attributes.at("linear_before_reset")).at<int>();
        }

Shucai Xiao's avatar
Shucai Xiao committed
936
        // append undefined opeator to make 6 arguments
Shucai Xiao's avatar
Shucai Xiao committed
937
        if(args.size() < 6)
Shucai Xiao's avatar
Shucai Xiao committed
938
939
940
941
942
        {
            auto ins = prog.add_instruction(op::undefined{});
            args.insert(args.end(), 6 - args.size(), ins);
        }

943
944
        // first output for concatenation of hidden states
        auto hidden_states = prog.add_instruction(
Shucai Xiao's avatar
Shucai Xiao committed
945
            op::gru{hidden_size, vec_actv_funcs, dirct, clip, linear_before_reset},
Shucai Xiao's avatar
Shucai Xiao committed
946
            std::move(args));
947
948

        // second output for last gru output
949
        auto last_output = prog.add_instruction(op::rnn_last_output{}, hidden_states);
950

Shucai Xiao's avatar
Shucai Xiao committed
951
        return {hidden_states, last_output};
952
953
    }

Shucai Xiao's avatar
Shucai Xiao committed
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
    std::vector<instruction_ref>
    parse_lstm(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        migraphx::shape input_shape = args[0]->get_shape();
        std::size_t hidden_size     = args[2]->get_shape().lens()[2];

        if(contains(attributes, "hidden_size"))
        {
            std::size_t hidden_size_att = parse_value(attributes.at("hidden_size")).at<int>();
            if(hidden_size != hidden_size_att)
            {
                MIGRAPHX_THROW("LSTM: hidden size mismatch in input and attribute");
            }
        }

        // Handling of direction to be added later
        std::string direction{"forward"};
        if(contains(attributes, "direction"))
        {
            direction = attributes.at("direction").s();
        }

Shucai Xiao's avatar
Shucai Xiao committed
976
        op::rnn_direction dirct = op::rnn_direction::forward;
Shucai Xiao's avatar
Shucai Xiao committed
977
978
        if(direction == "bidirectional")
        {
Shucai Xiao's avatar
Shucai Xiao committed
979
            dirct = op::rnn_direction::bidirectional;
Shucai Xiao's avatar
Shucai Xiao committed
980
981
982
        }
        else if(direction == "reverse")
        {
Shucai Xiao's avatar
Shucai Xiao committed
983
            dirct = op::rnn_direction::reverse;
Shucai Xiao's avatar
Shucai Xiao committed
984
        }
Shucai Xiao's avatar
Shucai Xiao committed
985
        else if(direction == "forward")
Shucai Xiao's avatar
Shucai Xiao committed
986
        {
Shucai Xiao's avatar
Shucai Xiao committed
987
            dirct = op::rnn_direction::forward;
Shucai Xiao's avatar
Shucai Xiao committed
988
989
990
991
992
993
994
995
996
997
998
999
        }
        else
        {
            MIGRAPHX_THROW("LSTM: incorrect direction attribute");
        }

        std::vector<std::string> vec_names = {"sigmoid", "tanh", "tanh"};
        if(contains(attributes, "activations"))
        {
            auto names = attributes.at("activations").strings();
            vec_names.clear();
            vec_names.resize(names.size());
1000
            std::copy(names.begin(), names.end(), vec_names.begin());
Shucai Xiao's avatar
Shucai Xiao committed
1001
1002
1003
        }

        // need 6 activation functions for bidirectional directions
Shucai Xiao's avatar
Shucai Xiao committed
1004
        if(dirct == op::rnn_direction::bidirectional)
Shucai Xiao's avatar
Shucai Xiao committed
1005
1006
1007
1008
1009
1010
        {
            // 6 activation functions are used in the bidirectional
            // scenario. No spec is provided in onnx::operator. we
            // use the algorithm that: if 1 actv function is provided,
            // repeat 1st six times. If 2 actv functins are provided,
            // repeat 2nd once, then repeat all three once
Shucai Xiao's avatar
Shucai Xiao committed
1011
            // if 3 actv funcs are provide, repeat all three once.
Shucai Xiao's avatar
Shucai Xiao committed
1012
1013
1014
1015
            // the same algorithm is used for 4, 5, and 6 actv funcions
            // provided. This may need change later
            switch(vec_names.size())
            {
1016
            case 1:
Shucai Xiao's avatar
Shucai Xiao committed
1017
1018
1019
1020
1021
1022
                vec_names = {vec_names.at(0),
                             vec_names.at(0),
                             vec_names.at(0),
                             vec_names.at(0),
                             vec_names.at(0),
                             vec_names.at(0)};
1023
                break;
Shucai Xiao's avatar
Shucai Xiao committed
1024
1025
1026

            case 2:
                // repeat the 2nd actv func once, then repeat all three another time
Shucai Xiao's avatar
Shucai Xiao committed
1027
1028
1029
1030
1031
1032
                vec_names = {vec_names.at(0),
                             vec_names.at(1),
                             vec_names.at(1),
                             vec_names.at(0),
                             vec_names.at(1),
                             vec_names.at(1)};
Shucai Xiao's avatar
Shucai Xiao committed
1033
1034
1035
1036
                break;

            case 3:
                // repeat all three actv funcs once
Shucai Xiao's avatar
Shucai Xiao committed
1037
1038
1039
1040
1041
1042
                vec_names = {vec_names.at(0),
                             vec_names.at(1),
                             vec_names.at(2),
                             vec_names.at(0),
                             vec_names.at(1),
                             vec_names.at(2)};
Shucai Xiao's avatar
Shucai Xiao committed
1043
1044
                break;

Shucai Xiao's avatar
Shucai Xiao committed
1045
1046
1047
1048
1049
1050
1051
            case 4:
                vec_names = {vec_names.at(0),
                             vec_names.at(1),
                             vec_names.at(2),
                             vec_names.at(3),
                             vec_names.at(3),
                             vec_names.at(3)};
1052
                break;
Shucai Xiao's avatar
Shucai Xiao committed
1053

Shucai Xiao's avatar
Shucai Xiao committed
1054
1055
1056
1057
1058
1059
1060
            case 5:
                vec_names = {vec_names.at(0),
                             vec_names.at(1),
                             vec_names.at(2),
                             vec_names.at(3),
                             vec_names.at(4),
                             vec_names.at(4)};
1061
                break;
Shucai Xiao's avatar
Shucai Xiao committed
1062

Shucai Xiao's avatar
Shucai Xiao committed
1063
            default: break;
Shucai Xiao's avatar
Shucai Xiao committed
1064
1065
1066
1067
1068
1069
            }
        }
        else
        {
            switch(vec_names.size())
            {
Shucai Xiao's avatar
Shucai Xiao committed
1070
            case 1: vec_names = {vec_names.at(0), vec_names.at(0), vec_names.at(0)}; break;
Shucai Xiao's avatar
Shucai Xiao committed
1071
1072
1073

            case 2:
                // repeat the 2nd actv func once, so we have 3 actv funcs
Shucai Xiao's avatar
Shucai Xiao committed
1074
                vec_names = {vec_names.at(0), vec_names.at(1), vec_names.at(1)};
Shucai Xiao's avatar
Shucai Xiao committed
1075
1076
                break;

Shucai Xiao's avatar
Shucai Xiao committed
1077
            default: break;
Shucai Xiao's avatar
Shucai Xiao committed
1078
1079
1080
            }
        }

1081
1082
1083
        auto name_it = std::find_if(vec_names.begin(), vec_names.end(), [&](auto& name) {
            return (map_actv_funcs.count(name) == 0);
        });
Shucai Xiao's avatar
Shucai Xiao committed
1084
        if(name_it != vec_names.end())
1085
1086
1087
        {
            MIGRAPHX_THROW("LSTM: activation function " + std::string(*name_it) + " not supported");
        }
Shucai Xiao's avatar
Shucai Xiao committed
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109

        std::vector<operation> vec_actv_funcs(vec_names.size());
        std::transform(vec_names.begin(), vec_names.end(), vec_actv_funcs.begin(), [&](auto& name) {
            return map_actv_funcs[name];
        });

        float clip = 0.0;
        if(contains(attributes, "clip"))
        {
            clip = parse_value(attributes.at("clip")).at<float>();
        }

        int input_forget = 0;
        if(contains(attributes, "input_forget"))
        {
            input_forget = parse_value(attributes.at("input_forget")).at<int>();
        }

        // append undefined opeator to make 6 arguments
        if(args.size() < 8)
        {
            auto ins = prog.add_instruction(op::undefined{});
Shucai Xiao's avatar
Shucai Xiao committed
1110
            args.insert(args.end(), 8 - args.size(), ins);
Shucai Xiao's avatar
Shucai Xiao committed
1111
1112
1113
1114
        }

        // first output for concatenation of hidden states
        auto hidden_states = prog.add_instruction(
Shucai Xiao's avatar
Shucai Xiao committed
1115
            op::lstm{hidden_size, vec_actv_funcs, dirct, clip, input_forget}, std::move(args));
Shucai Xiao's avatar
Shucai Xiao committed
1116
1117

        // second output for last lstm output
Shucai Xiao's avatar
Shucai Xiao committed
1118
        auto last_output = prog.add_instruction(op::rnn_last_output{}, hidden_states);
Shucai Xiao's avatar
Shucai Xiao committed
1119
1120
1121
1122
1123
1124
1125

        // third output for last cell output
        auto last_cell_output = prog.add_instruction(op::lstm_last_cell_output{}, hidden_states);

        return {hidden_states, last_output, last_cell_output};
    }

Paul's avatar
Paul committed
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
    void parse_from(std::istream& is)
    {
        onnx::ModelProto model;
        if(model.ParseFromIstream(&is))
        {
            if(model.has_graph())
            {
                this->parse_graph(model.graph());
            }
        }
        else
        {
Paul's avatar
Paul committed
1138
            MIGRAPHX_THROW("Failed reading onnx file.");
Paul's avatar
Paul committed
1139
1140
1141
1142
1143
1144
        }
    }

    void parse_graph(const onnx::GraphProto& graph)
    {
        nodes = get_nodes(graph);
1145
1146
1147
1148
1149
        std::unordered_map<std::string, onnx::TensorProto> initializer_data;
        for(auto&& f : graph.initializer())
        {
            initializer_data[f.name()] = f;
        }
Paul's avatar
Paul committed
1150
1151
1152
        for(auto&& input : graph.input())
        {
            const std::string& name = input.name();
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
            // Does the input have an initializer?
            if(contains(initializer_data, name))
            {
                auto t             = initializer_data[name];
                instructions[name] = prog.add_literal(parse_tensor(t));
            }
            else
            {
                // TODO: Get shape of input parameter
                shape s            = parse_type(input.type());
                instructions[name] = prog.add_parameter(name, s);
            }
Paul's avatar
Paul committed
1165
        }
Paul's avatar
Paul committed
1166
        for(auto&& output : graph.output())
Paul's avatar
Paul committed
1167
        {
Paul's avatar
Paul committed
1168
            this->parse_node(output.name());
Paul's avatar
Paul committed
1169
1170
1171
        }
    }

Shucai Xiao's avatar
Shucai Xiao committed
1172
    void parse_undefined(const std::string& name)
1173
    {
Shucai Xiao's avatar
Shucai Xiao committed
1174
        auto ins           = prog.add_instruction(op::undefined{});
1175
1176
1177
        instructions[name] = ins;
    }

Paul's avatar
Paul committed
1178
    void parse_node(const std::string& name)
Paul's avatar
Paul committed
1179
    {
Paul's avatar
Paul committed
1180
        if(name.empty())
Paul's avatar
Paul committed
1181
            MIGRAPHX_THROW("Onnx node must have a name");
Paul's avatar
Paul committed
1182
1183
1184
1185
1186
1187
1188
1189
        if(instructions.count(name) == 0)
        {
            auto&& node = nodes.at(name);
            std::vector<instruction_ref> args;
            for(auto&& input : node.input())
            {
                if(nodes.count(input) > 0)
                {
Paul's avatar
Paul committed
1190
1191
                    assert(name != input);
                    this->parse_node(input);
Paul's avatar
Paul committed
1192
                }
Shucai Xiao's avatar
Shucai Xiao committed
1193
                else if(input.empty())
Paul's avatar
Paul committed
1194
                {
1195
                    this->parse_undefined(input);
Paul's avatar
Paul committed
1196
                }
1197
                args.push_back(instructions.at(input));
Paul's avatar
Paul committed
1198
            }
Paul's avatar
Paul committed
1199
            std::vector<instruction_ref> result;
Paul's avatar
Paul committed
1200
1201
            if(ops.count(node.op_type()) == 0)
            {
Paul's avatar
Paul committed
1202
                result.push_back(prog.add_instruction(unknown{node.op_type()}, args));
Paul's avatar
Paul committed
1203
1204
1205
            }
            else
            {
Paul's avatar
Paul committed
1206
                result = ops[node.op_type()](get_attributes(node), args);
Paul's avatar
Paul committed
1207
            }
Paul's avatar
Paul committed
1208
            // Even no output nodes produce output in migraphx
Paul's avatar
Paul committed
1209
            if(node.output().empty() and result.size() == 1)
Paul's avatar
Paul committed
1210
1211
            {
                instructions[name] = result.front();
Paul's avatar
Paul committed
1212
1213
1214
            }
            else
            {
Paul's avatar
Paul committed
1215
1216
1217
1218
1219
1220
                assert(node.output().size() >= result.size());
                std::transform(result.begin(),
                               result.end(),
                               node.output().begin(),
                               std::inserter(instructions, instructions.end()),
                               [](auto&& x, auto&& y) { return std::make_pair(y, x); });
Paul's avatar
Paul committed
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
            }
        }
    }

    static attribute_map get_attributes(const onnx::NodeProto& node)
    {
        std::unordered_map<std::string, onnx::AttributeProto> result;
        for(auto&& attr : node.attribute())
        {
            result[attr.name()] = attr;
        }
        return result;
    }

    static node_map get_nodes(const onnx::GraphProto& graph)
    {
        std::unordered_map<std::string, onnx::NodeProto> result;
Paul's avatar
Paul committed
1238
        std::size_t n = 0;
Paul's avatar
Paul committed
1239
1240
        for(auto&& node : graph.node())
        {
Paul's avatar
Paul committed
1241
            if(node.output().empty())
Paul's avatar
Paul committed
1242
            {
Paul's avatar
Paul committed
1243
                if(node.name().empty())
Paul's avatar
Paul committed
1244
1245
1246
1247
1248
1249
1250
1251
1252
                {
                    result["migraphx_unamed_node_" + std::to_string(n)] = node;
                    n++;
                }
                else
                {
                    result[node.name()] = node;
                }
            }
Paul's avatar
Paul committed
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
            for(auto&& output : node.output())
            {
                result[output] = node;
            }
        }
        return result;
    }

    template <class T>
    static literal from_repeated(shape::type_t t, const T& r)
    {
        std::size_t size = r.size();
        return literal{{t, {size}}, r.begin(), r.end()};
    }

    static literal parse_value(const onnx::AttributeProto& attr)
    {
        switch(attr.type())
        {
        case onnx::AttributeProto::UNDEFINED: return {};
        case onnx::AttributeProto::FLOAT: return literal{attr.f()};
        case onnx::AttributeProto::INT: return literal{attr.i()};
        case onnx::AttributeProto::STRING: return {};
        case onnx::AttributeProto::TENSOR: return parse_tensor(attr.t());
        case onnx::AttributeProto::GRAPH: return {};
Paul's avatar
Paul committed
1278
        case onnx::AttributeProto::FLOATS: return from_repeated(shape::float_type, attr.floats());
Paul's avatar
Paul committed
1279
1280
1281
1282
1283
        case onnx::AttributeProto::INTS: return from_repeated(shape::int64_type, attr.ints());
        case onnx::AttributeProto::STRINGS: return {};
        case onnx::AttributeProto::TENSORS: return {};
        case onnx::AttributeProto::GRAPHS: return {};
        }
Paul's avatar
Paul committed
1284
        MIGRAPHX_THROW("Invalid attribute type");
Paul's avatar
Paul committed
1285
1286
1287
1288
1289
    }

    static literal parse_tensor(const onnx::TensorProto& t)
    {
        std::vector<std::size_t> dims(t.dims().begin(), t.dims().end());
Khalique's avatar
Khalique committed
1290
        // in case of scalar constants in onnx file, use dims=1 to fill initializer data
1291
        if(dims.empty())
Khalique's avatar
Khalique committed
1292
1293
1294
        {
            dims = {1};
        }
1295
1296
        if(t.has_raw_data())
        {
wsttiger's avatar
wsttiger committed
1297
            const std::string& s = t.raw_data();
Scott Thornton's avatar
Scott Thornton committed
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
            switch(t.data_type())
            {
            case onnx::TensorProto::UNDEFINED: throw std::runtime_error("");
            case onnx::TensorProto::FLOAT: return literal{{shape::float_type, dims}, s.data()};
            case onnx::TensorProto::UINT8: throw std::runtime_error("");
            case onnx::TensorProto::INT8: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::UINT16: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::INT16: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::INT32: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::INT64: return literal{{shape::int64_type, dims}, s.data()};
            case onnx::TensorProto::STRING: throw std::runtime_error("");
            case onnx::TensorProto::BOOL: return literal{{shape::int32_type, dims}, s.data()};
Paul's avatar
Paul committed
1310
            case onnx::TensorProto::FLOAT16: return literal{{shape::half_type, dims}, s.data()};
Scott Thornton's avatar
Scott Thornton committed
1311
1312
1313
1314
1315
1316
            case onnx::TensorProto::DOUBLE: return literal{{shape::double_type, dims}, s.data()};
            case onnx::TensorProto::UINT32: throw std::runtime_error("");
            case onnx::TensorProto::UINT64: throw std::runtime_error("");
            case onnx::TensorProto::COMPLEX64: throw std::runtime_error("");
            case onnx::TensorProto::COMPLEX128: throw std::runtime_error("");
            }
Paul's avatar
Paul committed
1317
            MIGRAPHX_THROW("Invalid tensor type");
1318
        }
Paul's avatar
Paul committed
1319
1320
1321
1322
        switch(t.data_type())
        {
        case onnx::TensorProto::UNDEFINED: throw std::runtime_error("");
        case onnx::TensorProto::FLOAT:
Paul's avatar
Paul committed
1323
            return literal{{shape::float_type, dims}, t.float_data().begin(), t.float_data().end()};
Paul's avatar
Paul committed
1324
1325
        case onnx::TensorProto::UINT8: throw std::runtime_error("");
        case onnx::TensorProto::INT8:
Paul's avatar
Paul committed
1326
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
1327
        case onnx::TensorProto::UINT16:
Paul's avatar
Paul committed
1328
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
1329
        case onnx::TensorProto::INT16:
Paul's avatar
Paul committed
1330
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
1331
        case onnx::TensorProto::INT32:
Paul's avatar
Paul committed
1332
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
1333
        case onnx::TensorProto::INT64:
Paul's avatar
Paul committed
1334
            return literal{{shape::int64_type, dims}, t.int64_data().begin(), t.int64_data().end()};
Paul's avatar
Paul committed
1335
1336
        case onnx::TensorProto::STRING: throw std::runtime_error("");
        case onnx::TensorProto::BOOL:
Paul's avatar
Paul committed
1337
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
1338
        case onnx::TensorProto::FLOAT16:
Khalique's avatar
Khalique committed
1339
        {
Khalique's avatar
Khalique committed
1340
            std::vector<uint16_t> data_uint16(t.int32_data().begin(), t.int32_data().end());
1341
            std::vector<half> data_half;
Khalique's avatar
Khalique committed
1342
1343
1344
            std::transform(data_uint16.begin(),
                           data_uint16.end(),
                           std::back_inserter(data_half),
1345
                           [](uint16_t raw_val) { return *reinterpret_cast<half*>(&raw_val); });
1346
            return literal{{shape::half_type, dims}, data_half.begin(), data_half.end()};
Khalique's avatar
Khalique committed
1347
        }
Paul's avatar
Paul committed
1348
1349
1350
1351
1352
1353
1354
1355
        case onnx::TensorProto::DOUBLE:
            return literal{
                {shape::double_type, dims}, t.double_data().begin(), t.double_data().end()};
        case onnx::TensorProto::UINT32: throw std::runtime_error("");
        case onnx::TensorProto::UINT64: throw std::runtime_error("");
        case onnx::TensorProto::COMPLEX64: throw std::runtime_error("");
        case onnx::TensorProto::COMPLEX128: throw std::runtime_error("");
        }
Paul's avatar
Paul committed
1356
        MIGRAPHX_THROW("Invalid tensor type");
Paul's avatar
Paul committed
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
    }

    static shape parse_type(const onnx::TypeProto& t)
    {
        shape::type_t shape_type{};
        switch(t.tensor_type().elem_type())
        {
        case onnx::TensorProto::UNDEFINED:
            break; // throw std::runtime_error("Unsupported type UNDEFINED");
        case onnx::TensorProto::FLOAT: shape_type = shape::float_type; break;
        case onnx::TensorProto::UINT8:
            break; // throw std::runtime_error("Unsupported type UINT8");
        case onnx::TensorProto::INT8: shape_type = shape::int8_type; break;
        case onnx::TensorProto::UINT16: shape_type = shape::uint16_type; break;
        case onnx::TensorProto::INT16: shape_type = shape::int16_type; break;
        case onnx::TensorProto::INT32: shape_type = shape::int32_type; break;
        case onnx::TensorProto::INT64: shape_type = shape::int64_type; break;
        case onnx::TensorProto::STRING:
            break; // throw std::runtime_error("Unsupported type STRING");
        case onnx::TensorProto::BOOL:
            break; // throw std::runtime_error("Unsupported type BOOL");
Paul's avatar
Paul committed
1378
        case onnx::TensorProto::FLOAT16: shape_type = shape::half_type; break;
Paul's avatar
Paul committed
1379
1380
1381
1382
1383
1384
1385
1386
1387
        case onnx::TensorProto::DOUBLE: shape_type = shape::double_type; break;
        case onnx::TensorProto::UINT32: shape_type = shape::uint32_type; break;
        case onnx::TensorProto::UINT64: shape_type = shape::uint64_type; break;
        case onnx::TensorProto::COMPLEX64:
            break; // throw std::runtime_error("Unsupported type COMPLEX64");
        case onnx::TensorProto::COMPLEX128:
            break; // throw std::runtime_error("Unsupported type COMPLEX128");
        }
        std::vector<std::size_t> dims;
Paul's avatar
Paul committed
1388
        auto&& tensor_dims = t.tensor_type().shape().dim();
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
        std::transform(tensor_dims.begin(),
                       tensor_dims.end(),
                       std::back_inserter(dims),
                       [](auto&& d) -> std::size_t {
                           if(not d.has_dim_value())
                           {
                               long default_batch_size = 1; // FIXME
                               return default_batch_size;
                           }
                           return d.dim_value();
                       });
Paul's avatar
Paul committed
1400
1401
        return {shape_type, dims};
    }
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423

    shape::type_t get_type(int dtype)
    {
        switch(dtype)
        {
        case 1: return shape::float_type;
        case 2: return shape::uint8_type;
        case 3: return shape::int8_type;
        case 4: return shape::uint16_type;
        case 5: return shape::int16_type;
        case 6: return shape::int32_type;
        case 7: return shape::int64_type;
        case 10: return shape::half_type;
        case 11: return shape::double_type;
        case 12: return shape::uint32_type;
        case 13: return shape::uint64_type;
        default:
        {
            MIGRAPHX_THROW("Prototensor data type " + std::to_string(dtype) + " not supported");
        }
        }
    }
Paul's avatar
Paul committed
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
};

program parse_onnx(const std::string& name)
{
    std::fstream input(name.c_str(), std::ios::in | std::ios::binary);
    onnx_parser parser;
#ifndef NDEBUG
    // Log the program when it can't be parsed
    try
    {
        parser.parse_from(input);
    }
    catch(...)
    {
        std::cerr << parser.prog << std::endl;
        throw;
    }
#else
    parser.parse_from(input);
#endif
    return std::move(parser.prog);
}

Paul's avatar
Paul committed
1447
} // namespace MIGRAPHX_INLINE_NS
Paul's avatar
Paul committed
1448
} // namespace migraphx