search

jacoxu / HMN4QA

This is the code&dataset for the paper [Hierarchical Memory Networks for Answer Selection on Unknown Words. COLING 2016]
45 stars 16 forks source link

3 errors detected in the compilation #1

Open SeekPoint opened 7 years ago

SeekPoint commented 7 years ago

rzai@rzai00:~/prj/HMN4QA$ CUDA_VISIBLE_DEVICES=1 python main_run.py Using gpu device 0: GeForce GTX 1080 (CNMeM is disabled) Go to model

Start to prepare data set

Test sample 0 : stories [22016 22015 22014 22013 22012 22011 22010 22009 22008 22007 22006 22005 22004 22003 22002 22001] Test sample 0 : queries 22017 Test sample 0 : answers 滕俊泽

Test sample 1 : stories [22016 22015 22014 22013 22012 22011 22010 22009 22008 22007 22006 22005 22004 22003 22002 22001] Test sample 1 : queries 22018 Test sample 1 : answers IST

Test sample 2 : stories [22016 22015 22014 22013 22012 22011 22010 22009 22008 22007 22006 22005 22004 22003 22002 22001] Test sample 2 : queries 22019 Test sample 2 : answers 南京

batch_size: 32 max_seq_story: 16 max_sent_enc: 17 all_text_idx: (44001, 17) vocab size: 8540 train stories (5400, 16) queries (5400,) answers (5400,) dev stories (600, 16) queries (600,) answers (600,) test stories (6000, 16) queries (6000,) answers (6000,) /usr/local/lib/python2.7/dist-packages/theano/scan_module/scan.py:1019: Warning: In the strict mode, all neccessary shared variables must be passed as a part of non_sequences 'must be passed as a part of non_sequences', Warning) params_sent_mem: [B_embedding.W, A_time0.T, C_embedding0.W, C_time0.T, C_embedding1.W, C_time1.T, C_embedding2.W, C_time2.T] params_word_mem: [B_embedding.W, A_time0.T, C_embedding0.W, C_time0.T, C_embedding1.W, C_time1.T, C_embedding2.W, W_in_to_updategate, W_hid_to_updategate, b_updategate, W_in_to_resetgate, W_hid_to_resetgate, b_resetgate, W_in_to_hidden_update, W_hid_to_hidden_update, b_hidden_update, hid_init, W_in_to_updategate, W_hid_to_updategate, b_updategate, W_in_to_resetgate, W_hid_to_resetgate, b_resetgate, W_in_to_hidden_update, W_hid_to_hidden_update, b_hidden_update, hid_init, word_mem_W_align, word_mem_U_align, word_mem_v_align] Start to compile, the process may cost tens of minutes ... 1 #include 2 #include 3 #include "theano_mod_helper.h" 4 #include "cuda_ndarray.cuh" 5 #include "cudnn.h" 6 #include "cudnn_helper.h" 7 ////////////////////// 8 //// Support Code 9 ////////////////////// 10 11 12 static cudnnHandle_t _handle = NULL; 13 14 15 static int 16 c_set_tensorNd(CudaNdarray var, cudnnTensorDescriptor_t desc) { 17 18 19 int dim = CudaNdarray_NDIM(var); 20 int strides[dim]; 21 int default_str = 1; 22 23 for (int i = dim-1; i >= 0; i--) 24 { 25 if (CudaNdarray_HOST_STRIDES(var)[i]) 26 strides[i] = CudaNdarray_HOST_STRIDES(var)[i]; 27 else 28 strides[i] = default_str; 29 default_str = CudaNdarray_HOST_DIMS(var)[i]; 30 } 31 32 cudnnStatus_t err = cudnnSetTensorNdDescriptor(desc, CUDNN_DATA_FLOAT, dim, 33 CudaNdarray_HOST_DIMS(var), 34 strides); 35 if (err != CUDNN_STATUS_SUCCESS) { 36 PyErr_Format(PyExc_RuntimeError, 37 "Could not set tensorNd descriptor: %s" 38 "dim=%d", 39 cudnnGetErrorString(err), dim); 40 return -1; 41 } 42 return 0; 43 } 44 45 46 static int 47 c_set_filterNd(CudaNdarray var, cudnnFilterDescriptor_t desc) { 48 if (!CudaNdarray_is_c_contiguous(var)) { 49 PyErr_SetString(PyExc_ValueError, 50 "Only contiguous filters (kernels) are supported."); 51 return -1; 52 } 53 int dim = CudaNdarray_NDIM(var); 54 cudnnStatus_t err = cudnnSetFilterNdDescriptor(desc, CUDNN_DATA_FLOAT, dim, 55 CudaNdarray_HOST_DIMS(var)); 56 if (err != CUDNN_STATUS_SUCCESS) { 57 PyErr_Format(PyExc_RuntimeError, 58 "Could not set filter descriptor: %s." 59 " dims= %d", 60 cudnnGetErrorString(err), dim); 61 return -1; 62 } 63 return 0; 64 } 65 66 67 68 namespace { 69 struct __struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e { 70 PyObject ERROR; 71 72 PyObject storage_V3; 73 PyObject storage_V1; 74
75 cudnnTensorDescriptor_t softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0; 76 77 cudnnTensorDescriptor_t softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0; 78 79 80 struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e() {} 81 ~struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e(void) { 82 cleanup(); 83 } 84 85 int init(PyObject __ERROR, PyObject storage_V3, PyObject* storage_V1) { 86 Py_XINCREF(storage_V3); 87 Py_XINCREF(storage_V1); 88 this->storage_V3 = storage_V3; 89 this->storage_V1 = storage_V1; 90
91 92 93 94 cudnnStatus_t errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0; 95 96 softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = NULL; 97 if ((errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = cudnnCreateTensorDescriptor(&softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0)) != CUDNN_STATUS_SUCCESS) { 98 PyErr_Format(PyExc_MemoryError, "could not allocate tensor descriptor " 99 ": %s", cudnnGetErrorString(errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0)); 100 { 101 if (!PyErr_Occurred()) { 102 PyErr_SetString(PyExc_RuntimeError, 103 "Unexpected error in an Op's C code. " 104 "No Python exception was set."); 105 } 106 return 5; 107 } 108 } 109 110 softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = NULL; 111 if ((errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = cudnnCreateTensorDescriptor(&softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0)) != CUDNN_STATUS_SUCCESS) { 112 PyErr_Format(PyExc_MemoryError, "could not allocate tensor descriptor " 113 ": %s", cudnnGetErrorString(errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0)); 114 { 115 if (!PyErr_Occurred()) { 116 PyErr_SetString(PyExc_RuntimeError, 117 "Unexpected error in an Op's C code. " 118 "No Python exception was set."); 119 } 120 return 5; 121 } 122 } 123 124 this->ERROR = ERROR; 125 return 0; 126 } 127 void cleanup(void) { 128 label_1: 129 130 double DUMMY_1; 131 __label_3: 132 133 double DUMMY_3; 134 label_6: 135 136 if(softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0!= NULL) 137 cudnnDestroyTensorDescriptor(softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0); 138 139 if(softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0!= NULL) 140 cudnnDestroyTensorDescriptor(softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0); 141 142 double __DUMMY_6; 143 144 Py_XDECREF(this->storage_V3); 145 Py_XDECREF(this->storage_V1); 146 } 147 int run(void) { 148 int failure = 0; 149
150 PyObject py_V1; 151 CudaNdarray V1; 152 PyObject py_V3; 153 CudaNdarray V3; 154 { 155 156 py_V1 = PyList_GET_ITEM(storage_V1, 0); 157 {Py_XINCREF(py_V1);} 158
159 if (py_V1 == Py_None) 160 { 161 V1 = NULL; 162 } 163 else 164 { 165
166 assert(py_V1->ob_refcnt >= 2); // There should be at least one ref from the container object, 167 // and one ref from the local scope. 168 169 if (CudaNdarray_Check(py_V1)) 170 { 171 //fprintf(stderr, "c_extract CNDA object w refcnt %p %i\n", py_V1, (py_V1->ob_refcnt)); 172 V1 = (CudaNdarray)py_V1; 173 //std::cerr << "c_extract " << V1 << '\n'; 174
175 176 if (V1->nd != 4) 177 { 178 PyErr_Format(PyExc_RuntimeError, 179 "c_extract: Some CudaNdarray has rank %i, it was supposed to have rank 4", 180 V1->nd); 181 V1 = NULL; 182 { 183 failure = 2; 184 if (!PyErr_Occurred()) { 185 PyErr_SetString(PyExc_RuntimeError, 186 "Unexpected error in an Op's C code. " 187 "No Python exception was set."); 188 } 189 goto label_2;}; 190 } 191 //std::cerr << "c_extract " << V1 << " nd check passed\n"; 192
193 194 assert(V1); 195 Py_INCREF(py_V1); 196 } 197 else if (py_V1 == Py_None) 198 { 199 PyErr_SetString(PyExc_TypeError, 200 "expected a CudaNdarray, not None"); 201 V1 = NULL; 202 { 203 failure = 2; 204 if (!PyErr_Occurred()) { 205 PyErr_SetString(PyExc_RuntimeError, 206 "Unexpected error in an Op's C code. " 207 "No Python exception was set."); 208 } 209 goto label_2;}; 210 } 211 else 212 { 213 //fprintf(stderr, "FAILING c_extract CNDA object w refcnt %p %i\n", py_V1, (py_V1->ob_refcnt)); 214 PyErr_SetString(PyExc_TypeError, "Argument not a CudaNdarray"); 215 V1 = NULL; 216 { 217 __failure = 2; 218 if (!PyErr_Occurred()) { 219 PyErr_SetString(PyExc_RuntimeError, 220 "Unexpected error in an Op's C code. " 221 "No Python exception was set."); 222 } 223 goto __label_2;}; 224 } 225 //std::cerr << "c_extract done " << V1 << '\n'; 226
227 228 } 229
230 { 231 232 py_V3 = PyList_GET_ITEM(storage_V3, 0); 233 {Py_XINCREF(py_V3);} 234
235 assert(py_V3->ob_refcnt >= 2); // There should be at least one ref from the container object, 236 // and one ref from the local scope. 237 238 if (CudaNdarray_Check(py_V3)) 239 { 240 //fprintf(stderr, "c_extract CNDA object w refcnt %p %i\n", py_V3, (py_V3->ob_refcnt)); 241 V3 = (CudaNdarray)py_V3; 242 //std::cerr << "c_extract " << V3 << '\n'; 243
244 245 if (V3->nd != 4) 246 { 247 PyErr_Format(PyExc_RuntimeError, 248 "c_extract: Some CudaNdarray has rank %i, it was supposed to have rank 4", 249 V3->nd); 250 V3 = NULL; 251 { 252 failure = 4; 253 if (!PyErr_Occurred()) { 254 PyErr_SetString(PyExc_RuntimeError, 255 "Unexpected error in an Op's C code. " 256 "No Python exception was set."); 257 } 258 goto label_4;}; 259 } 260 //std::cerr << "c_extract " << V3 << " nd check passed\n"; 261
262 263 assert(V3); 264 Py_INCREF(py_V3); 265 } 266 else if (py_V3 == Py_None) 267 { 268 PyErr_SetString(PyExc_TypeError, 269 "expected a CudaNdarray, not None"); 270 V3 = NULL; 271 { 272 failure = 4; 273 if (!PyErr_Occurred()) { 274 PyErr_SetString(PyExc_RuntimeError, 275 "Unexpected error in an Op's C code. " 276 "No Python exception was set."); 277 } 278 goto label_4;}; 279 } 280 else 281 { 282 //fprintf(stderr, "FAILING c_extract CNDA object w refcnt %p %i\n", py_V3, (py_V3->ob_refcnt)); 283 PyErr_SetString(PyExc_TypeError, "Argument not a CudaNdarray"); 284 V3 = NULL; 285 { 286 failure = 4; 287 if (!PyErr_Occurred()) { 288 PyErr_SetString(PyExc_RuntimeError, 289 "Unexpected error in an Op's C code. " 290 "No Python exception was set."); 291 } 292 goto label_4;}; 293 } 294 //std::cerr << "c_extract done " << V3 << '\n'; 295
296 297 { 298 // Op class GpuDnnSoftmax 299 300 cudnnStatus_t errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0; 301 cudnnTensorFormat_t formatnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = CUDNN_TENSOR_NCHW; 302 if (0 == 1) 303 formatnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = CUDNN_TENSOR_NHWC; 304 305 cudnnSoftmaxAlgorithm_t algonode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = CUDNN_SOFTMAX_ACCURATE; 306 307 cudnnSoftmaxMode_t modenode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = CUDNN_SOFTMAX_MODE_CHANNEL; 308 if (0 == 1) 309 modenode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = CUDNN_SOFTMAX_MODE_INSTANCE; 310 311 { 312 int str0, str1, str2, str3; 313 str3 = CudaNdarray_HOST_STRIDES(V3)[3]?CudaNdarray_HOST_STRIDES(V3)[3]:1; 314 str2 = CudaNdarray_HOST_STRIDES(V3)[2]?CudaNdarray_HOST_STRIDES(V3)[2]:CudaNdarray_HOST_DIMS(V3)[3]; 315 str1 = CudaNdarray_HOST_STRIDES(V3)[1]?CudaNdarray_HOST_STRIDES(V3)[1]:CudaNdarray_HOST_DIMS(V3)[2]CudaNdarray_HOST_DIMS(V3)[3]; 316 str0 = CudaNdarray_HOST_STRIDES(V3)[0]?CudaNdarray_HOST_STRIDES(V3)[0]:CudaNdarray_HOST_DIMS(V3)[2]CudaNdarray_HOST_DIMS(V3)[3]*CudaNdarray_HOST_DIMS(V3)[1]; 317 errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = cudnnSetTensor4dDescriptorEx( 318 softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, CUDNN_DATA_FLOAT, 319 CudaNdarray_HOST_DIMS(V3)[0], 320 CudaNdarray_HOST_DIMS(V3)[1], 321 CudaNdarray_HOST_DIMS(V3)[2], 322 CudaNdarray_HOST_DIMS(V3)[3], 323 str0, str1, str2, str3 324 ); 325 if (errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 != CUDNN_STATUS_SUCCESS) { 326 PyErr_Format(PyExc_RuntimeError, 327 "could not set tensor4d descriptor: %s" 328 "shapes=%d %d %d %d strides=%d %d %d %d", 329 cudnnGetErrorString(errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0), 330 CudaNdarray_HOST_DIMS(V3)[0], 331 CudaNdarray_HOST_DIMS(V3)[1], 332 CudaNdarray_HOST_DIMS(V3)[2], 333 CudaNdarray_HOST_DIMS(V3)[3], 334 str0, str1, str2, str3 335 ); 336 { 337 failure = 5; 338 if (!PyErr_Occurred()) { 339 PyErr_SetString(PyExc_RuntimeError, 340 "Unexpected error in an Op's C code. " 341 "No Python exception was set."); 342 } 343 goto label_5;} 344 } 345 } 346 347
348 if (CudaNdarray_prep_output(&V1, 4, CudaNdarray_HOST_DIMS(V3)) != 0) 349 { 350 { 351 failure = 5; 352 if (!PyErr_Occurred()) { 353 PyErr_SetString(PyExc_RuntimeError, 354 "Unexpected error in an Op's C code. " 355 "No Python exception was set."); 356 } 357 goto __label_5;} 358 } 359 360 { 361 int str0, str1, str2, str3; 362 str3 = CudaNdarray_HOST_STRIDES(V1)[3]?CudaNdarray_HOST_STRIDES(V1)[3]:1; 363 str2 = CudaNdarray_HOST_STRIDES(V1)[2]?CudaNdarray_HOST_STRIDES(V1)[2]:CudaNdarray_HOST_DIMS(V1)[3]; 364 str1 = CudaNdarray_HOST_STRIDES(V1)[1]?CudaNdarray_HOST_STRIDES(V1)[1]:CudaNdarray_HOST_DIMS(V1)[2]CudaNdarray_HOST_DIMS(V1)[3]; 365 str0 = CudaNdarray_HOST_STRIDES(V1)[0]?CudaNdarray_HOST_STRIDES(V1)[0]:CudaNdarray_HOST_DIMS(V1)[2]CudaNdarray_HOST_DIMS(V1)[3]*CudaNdarray_HOST_DIMS(V1)[1]; 366 errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = cudnnSetTensor4dDescriptorEx( 367 softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, CUDNN_DATA_FLOAT, 368 CudaNdarray_HOST_DIMS(V1)[0], 369 CudaNdarray_HOST_DIMS(V1)[1], 370 CudaNdarray_HOST_DIMS(V1)[2], 371 CudaNdarray_HOST_DIMS(V1)[3], 372 str0, str1, str2, str3 373 ); 374 if (errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 != CUDNN_STATUS_SUCCESS) { 375 PyErr_Format(PyExc_RuntimeError, 376 "could not set tensor4d descriptor: %s" 377 "shapes=%d %d %d %d strides=%d %d %d %d", 378 cudnnGetErrorString(errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0), 379 CudaNdarray_HOST_DIMS(V1)[0], 380 CudaNdarray_HOST_DIMS(V1)[1], 381 CudaNdarray_HOST_DIMS(V1)[2], 382 CudaNdarray_HOST_DIMS(V1)[3], 383 str0, str1, str2, str3 384 ); 385 { 386 failure = 5; 387 if (!PyErr_Occurred()) { 388 PyErr_SetString(PyExc_RuntimeError, 389 "Unexpected error in an Op's C code. " 390 "No Python exception was set."); 391 } 392 goto label_5;} 393 } 394 } 395 396
397 #ifndef CUDNN_VERSION 398 errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = cudnnSoftmaxForward( 399 _handle, 400 algonode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 401 modenode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 402 softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 403 CudaNdarray_DEV_DATA(V3), 404 softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 405 CudaNdarray_DEV_DATA(V1) 406 ); 407 #else 408 { 409 const float alpha = 1.; 410 const float beta = 0.; 411 errnode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0 = cudnnSoftmaxForward( 412 _handle, 413 algonode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 414 modenode_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 415 (void) &alpha, 416 softmax_input_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 417 CudaNdarray_DEV_DATA(V3), 418 (void) &beta, 419 softmax_output_node_e4b7b2bc7fd903d81ee9c46ba97aa65e_0, 420 CudaNdarray_DEV_DATA(V1) 421 ); 422 } 423 #endif 424 label_5: 425 426 double DUMMY_5; 427 428 } 429 label_4: 430 431 //std::cerr << "cleanup " << py_V3 << " " << V3 << "\n"; 432 //fprintf(stderr, "c_cleanup CNDA py_object w refcnt %p %i\n", py_V3, (py_V3->ob_refcnt)); 433 if (V3) 434 { 435 //fprintf(stderr, "c_cleanup CNDA cn_object w refcnt %p %i\n", V3, (V3->ob_refcnt)); 436 Py_XDECREF(V3); 437 } 438 //std::cerr << "cleanup done" << py_V3 << "\n"; 439
440 {Py_XDECREF(py_V3);} 441
442 double DUMMY_4; 443 444 } 445 label_2: 446 447 if (!failure) { 448
449 //std::cerr << "sync\n"; 450 if (NULL == V1) { 451 // failure: sync None to storage 452 Py_XDECREF(py_V1); 453 py_V1 = Py_None; 454 Py_INCREF(py_V1); 455 } 456 else 457 { 458 if (py_V1 != (PyObject)V1) 459 { 460 Py_XDECREF(py_V1); 461 py_V1 = (PyObject)V1; 462 Py_INCREF(py_V1); 463 } 464 assert(py_V1->ob_refcnt); 465 } 466
467 PyObject* old = PyList_GET_ITEM(storage_V1, 0); 468 {Py_XINCREF(py_V1);} 469 PyList_SET_ITEM(storage_V1, 0, py_V1); 470 {Py_XDECREF(old);} 471 } 472
473 //std::cerr << "cleanup " << py_V1 << " " << V1 << "\n"; 474 //fprintf(stderr, "c_cleanup CNDA py_object w refcnt %p %i\n", py_V1, (py_V1->ob_refcnt)); 475 if (V1) 476 { 477 //fprintf(stderr, "c_cleanup CNDA cn_object w refcnt %p %i\n", V1, (V1->ob_refcnt)); 478 Py_XDECREF(V1); 479 } 480 //std::cerr << "cleanup done" << py_V1 << "\n"; 481
482 {Py_XDECREF(py_V1);} 483
484 double __DUMMY_2; 485 486 } 487 488
489 if (failure) { 490 // When there is a failure, this code puts the exception 491 // in ERROR. 492 PyObject err_type = NULL; 493 PyObject err_msg = NULL; 494 PyObject err_traceback = NULL; 495 PyErr_Fetch(&err_type, &err_msg, &err_traceback); 496 if (!err_type) {err_type = Py_None;Py_INCREF(Py_None);} 497 if (!err_msg) {err_msg = Py_None; Py_INCREF(Py_None);} 498 if (!err_traceback) {err_traceback = Py_None; Py_INCREF(Py_None);} 499 PyObject old_err_type = PyList_GET_ITEM(ERROR, 0); 500 PyObject old_err_msg = PyList_GET_ITEM(__ERROR, 1); 501 PyObject old_err_traceback = PyList_GET_ITEM(ERROR, 2); 502 PyList_SET_ITEM(ERROR, 0, err_type); 503 PyList_SET_ITEM(ERROR, 1, err_msg); 504 PyList_SET_ITEM(ERROR, 2, err_traceback); 505 {Py_XDECREF(old_err_type);} 506 {Py_XDECREF(old_err_msg);} 507 {Py_XDECREF(old_err_traceback);} 508 } 509 // The failure code is returned to index what code block failed. 510 return failure; 511
512 } 513 }; 514 } 515
516 517 static int struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e_executor(struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e* self) { 518 return self->run(); 519 } 520 521 static void struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e_destructor(void executor, void self) { 522 delete ((struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e)self); 523 } 524
525 ////////////////////// 526 //// Functions 527 ////////////////////// 528 static PyObject instantiate(PyObject self, PyObject argtuple) { 529 assert(PyTuple_Check(argtuple)); 530 if (3 != PyTuple_Size(argtuple)){ 531 PyErr_Format(PyExc_TypeError, "Wrong number of arguments, expected 3, got %i", (int)PyTuple_Size(argtuple)); 532 return NULL; 533 } 534 struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e struct_ptr = new __struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e(); 535 if (struct_ptr->init( PyTuple_GET_ITEM(argtuple, 0),PyTuple_GET_ITEM(argtuple, 1),PyTuple_GET_ITEM(argtuple, 2) ) != 0) { 536 delete struct_ptr; 537 return NULL; 538 } 539 PyObject thunk = PyCObject_FromVoidPtrAndDesc((void*)(&__struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e_executor), struct_ptr, __struct_compiled_op_e4b7b2bc7fd903d81ee9c46ba97aa65e_destructor); 540 return thunk; } 541 542 ////////////////////// 543 //// Module init 544 ////////////////////// 545 static PyMethodDef MyMethods[] = { 546 {"instantiate", instantiate, METH_VARARGS, "undocumented"} , 547 {NULL, NULL, 0, NULL} 548 }; 549 PyMODINIT_FUNC inite4b7b2bc7fd903d81ee9c46ba97aa65e(void){ 550
551 552 { 553 cudnnStatus_t err; 554 if ((err = cudnnCreate(&_handle)) != CUDNN_STATUS_SUCCESS) { 555 PyErr_Format(PyExc_RuntimeError, "could not create cuDNN handle: %s", 556 cudnnGetErrorString(err)); 557 #if PY_MAJOR_VERSION >= 3 558 return NULL; 559 #else 560 return; 561 #endif 562 } 563 } 564 565 (void) Py_InitModule("e4b7b2bc7fd903d81ee9c46ba97aa65e", MyMethods); 566 } 567

mod.cu(54): error: argument of type "int" is incompatible with parameter of type "cudnnTensorFormat_t" mod.cu(55): error: argument of type "const int *" is incompatible with parameter of type "int" mod.cu(55): error: too few arguments in function call mod.cu(16): warning: function "c_set_tensorNd" was declared but never referenced mod.cu(47): warning: function "c_set_filterNd" was declared but never referenced 3 errors detected in the compilation of "/tmp/tmpxft_00004094_00000000-9_mod.cpp1.ii".

['nvcc', '-shared', '-O3', '-use_fast_math', '-arch=sm_61', '-m64', '-Xcompiler', '-fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=c72d035fdf91890f3b36710688069b2e,-DNPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-fPIC,-fvisibility=hidden', '-Xlinker', '-rpath,/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/cuda_ndarray', '-I/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/cuda_ndarray', '-I/usr/local/cuda-8.0/include', '-I/usr/local/lib/python2.7/dist-packages/theano/sandbox/cuda', '-I/home/rzai/.local/lib/python2.7/site-packages/numpy/core/include', '-I/usr/include/python2.7', '-I/usr/local/lib/python2.7/dist-packages/theano/gof', '-o', '/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/tmpQyHjmt/e4b7b2bc7fd903d81ee9c46ba97aa65e.so', 'mod.cu', '-L/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/cuda_ndarray', '-L/usr/lib', '-lpython2.7', '-lcudart', '-lcublas', '-lcuda_ndarray', '-lcudnn'] Traceback (most recent call last): File "main_run.py", line 31, in main() File "main_run.py", line 25, in main hmn4qa_model = model.Model(_log_file) File "/home/rzai/prj/HMN4QA/model.py", line 21, in init self.networks = self.build_network() File "/home/rzai/prj/HMN4QA/model.py", line 140, in build_network updates=updates_joint) File "/usr/local/lib/python2.7/dist-packages/theano/compile/function.py", line 316, in function output_keys=output_keys) File "/usr/local/lib/python2.7/dist-packages/theano/compile/pfunc.py", line 523, in pfunc output_keys=output_keys) File "/usr/local/lib/python2.7/dist-packages/theano/compile/function_module.py", line 1526, in orig_function defaults) File "/usr/local/lib/python2.7/dist-packages/theano/compile/function_module.py", line 1390, in create input_storage=input_storage_lists) File "/usr/local/lib/python2.7/dist-packages/theano/gof/link.py", line 607, in make_thunk output_storage=output_storage)[:3] File "/usr/local/lib/python2.7/dist-packages/theano/gof/vm.py", line 1025, in make_all no_recycling)) File "/usr/local/lib/python2.7/dist-packages/theano/sandbox/cuda/init.py", line 279, in make_thunk compute_map, no_recycling) File "/usr/local/lib/python2.7/dist-packages/theano/gof/op.py", line 807, in make_thunk no_recycling) File "/usr/local/lib/python2.7/dist-packages/theano/gof/op.py", line 733, in make_c_thunk output_storage=node_output_storage) File "/usr/local/lib/python2.7/dist-packages/theano/gof/cc.py", line 1065, in make_thunk keep_lock=keep_lock) File "/usr/local/lib/python2.7/dist-packages/theano/gof/cc.py", line 1007, in compile keep_lock=keep_lock) File "/usr/local/lib/python2.7/dist-packages/theano/gof/cc.py", line 1435, in cthunk_factory key=key, lnk=self, keep_lock=keep_lock) File "/usr/local/lib/python2.7/dist-packages/theano/gof/cmodule.py", line 1094, in module_from_key module = lnk.compile_cmodule(location) File "/usr/local/lib/python2.7/dist-packages/theano/gof/cc.py", line 1347, in compile_cmodule preargs=preargs) File "/usr/local/lib/python2.7/dist-packages/theano/sandbox/cuda/nvcc_compiler.py", line 429, in compile_str 'for cmd', ' '.join(cmd)) Exception: ('The following error happened while compiling the node', GpuDnnSoftmax{tensor_format='bc01', mode='channel', algo='accurate'}(GpuContiguous.0), '\n', 'nvcc return status', 2, 'for cmd', 'nvcc -shared -O3 -use_fast_math -arch=sm_61 -m64 -Xcompiler -fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=c72d035fdf91890f3b36710688069b2e,-DNPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-fPIC,-fvisibility=hidden -Xlinker -rpath,/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/cuda_ndarray -I/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/cuda_ndarray -I/usr/local/cuda-8.0/include -I/usr/local/lib/python2.7/dist-packages/theano/sandbox/cuda -I/home/rzai/.local/lib/python2.7/site-packages/numpy/core/include -I/usr/include/python2.7 -I/usr/local/lib/python2.7/dist-packages/theano/gof -o /home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/tmpQyHjmt/e4b7b2bc7fd903d81ee9c46ba97aa65e.so mod.cu -L/home/rzai/.theano/compiledir_Linux-4.4--generic-x86_64-with-Ubuntu-14.04-trusty-x86_64-2.7.6-64/cuda_ndarray -L/usr/lib -lpython2.7 -lcudart -lcublas -lcuda_ndarray -lcudnn', "[GpuDnnSoftmax{tensor_format='bc01', mode='channel', algo='accurate'}(<CudaNdarrayType(float32, (False, False, True, True))>)]") rzai@rzai00:~/prj/HMN4QA$

jacoxu commented 7 years ago

Sorry for that I have not encountered these errors. I think the most problem is that you use too newest CuDNN or some other package. Maybe you can find some clues from the following site:

https://groups.google.com/forum/#!topic/lasagne-users/CV2Gx8NZPQY

best wishes!