Closed mcsieber closed 3 years ago
Here is the markdown of the Jupyter Notebook to reproduce the problem:
coreml_str = '/Volumes/ArielD/storage/data/mlmodels/Resnet50/Resnet50.mlmodel' from coremltools import utils as cu from coremltools import models as cm
rn50 = cm.MLModel(coreml_str); type(rn50) coremltools.models.model.MLModel
rn50_spec = rn50.get_spec(); type(rn50_spec) Model_pb2.Model
rn50_loaded_spec = cu.load_spec(coreml_str); type(rn50_loaded_spec) Model_pb2.Model
#### Attempt to get NN Shaper crashes Python ```python rn50_shaper = cm.NeuralNetworkShaper(rn50_spec)
Process: python3.6 [3112] Path: /Users/USER/*/python Identifier: python3.6 Version: 0 Code Type: X86-64 (Native) Parent Process: python3.6 [2520] Responsible: python3.6 [3112] User ID: 501 Date/Time: 2019-07-20 17:08:35.308 -0400 OS Version: Mac OS X 10.14.5 (18F132) Report Version: 12 Anonymous UUID: DE8FAC1E-8FB2-7200-CE0B-B9CDBAC486EE Time Awake Since Boot: 26000 seconds System Integrity Protection: enabled Crashed Thread: 0 Dispatch queue: com.apple.main-thread Exception Type: EXC_CRASH (SIGABRT) Exception Codes: 0x0000000000000000, 0x0000000000000000 Exception Note: EXC_CORPSE_NOTIFY Application Specific Information: abort() called python(3112,0x1199e65c0) malloc: *** error for object 0x11374a008: pointer being freed was not allocated
(... long list of thread dumps and binary images ... )
VM Region Summary: ReadOnly portion of Libraries: Total=475.1M resident=0K(0%) swapped_out_or_unallocated=475.1M(100%) Writable regions: Total=951.7M written=0K(0%) resident=0K(0%) swapped_out=0K(0%) unallocated=951.7M(100%) VIRTUAL REGION REGION TYPE SIZE COUNT (non-coalesced) =========== ======= ======= Accelerate framework 128K 1 Activity Tracing 256K 1 CG backing stores 560K 3 CoreImage 8K 2 CoreUI image data 52K 1 CoreUI image file 404K 5 Dispatch continuations 8192K 1 Kernel Alloc Once 8K 1 MALLOC 869.9M 117 MALLOC guard page 32K 8 MALLOC_LARGE (reserved) 256K 2 reserved VM address space (unallocated) Memory Tag 242 12K 1 STACK GUARD 52K 13 Stack 39.7M 13 VM_ALLOCATE 96K 14 VM_ALLOCATE (reserved) 32.0M 1 reserved VM address space (unallocated) __DATA 27.3M 335 __DATA_CONST 68K 1 __FONT_DATA 4K 1 __LINKEDIT 235.0M 91 __TEXT 240.1M 328 __UNICODE 564K 1 mapped file 152.8M 17 shared memory 3244K 13 =========== ======= ======= TOTAL 1.6G 971 TOTAL, minus reserved VM space 1.5G 971 Model: MacBookPro9,2, BootROM 226.0.0.0.0, 2 processors, Intel Core i7, 2.9 GHz, 16 GB, SMC 2.2f44 Graphics: kHW_IntelHD4000Item, Intel HD Graphics 4000, spdisplays_builtin Memory Module: BANK 0/DIMM0, 8 GB, DDR3, 1600 MHz, 0x0000, - Memory Module: BANK 1/DIMM0, 8 GB, DDR3, 1600 MHz, 0x0000, - AirPort: spairport_wireless_card_type_airport_extreme (0x14E4, 0xF5), Broadcom BCM43xx 1.0 (7.21.190.33 AirPortDriverBrcm4360-1325.2) Bluetooth: Version 6.0.12f1, 3 services, 18 devices, 1 incoming serial ports Network Service: Ethernet, Ethernet, en0 Serial ATA Device: Mercury Electra 6G SSD, 1 TB Serial ATA Device: OPTIARC DVD RW AD-5970H USB Device: USB 2.0 Bus USB Device: Hub USB Device: FaceTime HD Camera (Built-in) USB Device: USB 2.0 Bus USB Device: Hub USB Device: Hub USB Device: Apple Internal Keyboard / Trackpad USB Device: IR Receiver USB Device: BRCM20702 Hub USB Device: Bluetooth USB Host Controller USB Device: Hub USB Device: Logitech USB Optical Mouse USB Device: Keyboard Hub USB Device: Apple Keyboard USB Device: USB 3.0 Bus Thunderbolt Bus: MacBook Pro, Apple Inc., 25.1
I got the same error on MacOS 10.15 with Python 2.7.14 and coremltools 3.03b.
The NeuralNetworkShaper has been removed from coremltools.
Here is the markdown of the Jupyter Notebook to reproduce the problem:
Coremltools 3.03b NeuralNetworkShaper Resnet50 crashes Python
Python 3.6.7 on MacOS 10.14.5
Create Model from Resnet50 model file
rn50_spec = rn50.get_spec(); type(rn50_spec) Model_pb2.Model
rn50_loaded_spec = cu.load_spec(coreml_str); type(rn50_loaded_spec) Model_pb2.Model
Excerpts from MacOS "Problem Report for Python"
(... long list of thread dumps and binary images ... )