Deepcopy of torchvision.models.feature_extraction.DualGraphModule breaks eval

🐛 Describe the bug

It appears that creating a deepcopy of a DualGraphModule breaks the ability to call .eval() on it.

Minimal example:

import torchvision
from torchvision.models import efficientnet as enet
import torchvision.models.feature_extraction
from copy import deepcopy

model = enet.efficientnet_b3(weights=enet.EfficientNet_B3_Weights.DEFAULT)
extractor = torchvision.models.feature_extraction.create_feature_extractor(model=model, return_nodes={"classifier.0": "out"})

extractor.eval()
extractor.train()

extractor_2 = deepcopy(extractor)
# the line below causes AttributeError: 'GraphModule' object has no attribute 'eval_graph'
extractor_2.eval()

Encountered this while creating an EMA of a model that had a module created with create_feature_extractor

Versions

Collecting environment information... PyTorch version: 2.4.0+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.4 LTS (x86_64) GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 Clang version: Could not collect CMake version: version 3.22.1 Libc version: glibc-2.35

Python version: 3.10.12 (main, Jul 29 2024, 16:56:48) [GCC 11.4.0] (64-bit runtime) Python platform: Linux-6.8.0-40-generic-x86_64-with-glibc2.35 Is CUDA available: True CUDA runtime version: Could not collect CUDA_MODULE_LOADING set to: LAZY GPU models and configuration: GPU 0: NVIDIA GeForce RTX 3090 Ti Nvidia driver version: 535.183.01 cuDNN version: Could not collect HIP runtime version: N/A MIOpen runtime version: N/A Is XNNPACK available: True

CPU: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 43 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 48 On-line CPU(s) list: 0-47 Vendor ID: AuthenticAMD Model name: AMD Ryzen Threadripper 3960X 24-Core Processor CPU family: 23 Model: 49 Thread(s) per core: 2 Core(s) per socket: 24 Socket(s): 1 Stepping: 0 Frequency boost: enabled CPU max MHz: 4568.1641 CPU min MHz: 2200.0000 BogoMIPS: 7585.83 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl umip rdpid overflow_recov succor smca sev sev_es Virtualization: AMD-V L1d cache: 768 KiB (24 instances) L1i cache: 768 KiB (24 instances) L2 cache: 12 MiB (24 instances) L3 cache: 128 MiB (8 instances) NUMA node(s): 1 NUMA node0 CPU(s): 0-47 Vulnerability Gather data sampling: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Mitigation; untrained return thunk; SMT enabled with STIBP protection Vulnerability Spec rstack overflow: Mitigation; Safe RET Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Retpolines; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected

Versions of relevant libraries: [pip3] denoising-diffusion-pytorch==1.11.0 [pip3] ema-pytorch==0.4.5 [pip3] mypy-extensions==1.0.0 [pip3] numpy==1.26.4 [pip3] pytorch-fid==0.3.0 [pip3] pytorch-lightning==2.2.1 [pip3] torch==2.4.0 [pip3] torchmetrics==1.3.2 [pip3] torchvision==0.19.0 [pip3] triton==3.0.0 [conda] Could not collect

pytorch / vision

Deepcopy of torchvision.models.feature_extraction.DualGraphModule breaks eval #8634

🐛 Describe the bug

Versions