Description
🐛 Describe the bug
I'm implementing an stft layer that is not trainable using 1d convolutions. One of the steps involves the normalization of a tensor. I'm finding that when I use the layer in a model, it only trains if I use in-place indexing.
I'm trying to divide one tensor by another in torch, but only when the values of the denominator exceed a certain threshold. This implementation works.
wsq_ola = wsq_ola.to(wav).expand_as(wav).clone()
min_mask = wsq_ola.abs() < eps
wav[~min_mask] = wav[~min_mask] / wsq_ola[~min_mask]I tried to implement the same thing with torch.where instead as follows:
wsq_ola = wsq_ola.to(wav).expand_as(wav).clone()
min_mask = wsq_ola.abs() < eps
wav = torch.where(min_mask, wav, wav / wsq_ola)Unfortunately, once I make this change, the model no longer learns. Is the gradient not propagated through torch.where?
Versions
PyTorch version: 1.13.0
Is debug build: False
CUDA used to build PyTorch: 11.7
ROCM used to build PyTorch: N/A
OS: Ubuntu 20.04.5 LTS (x86_64)
GCC version: (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
Clang version: Could not collect
CMake version: version 3.16.3
Libc version: glibc-2.31
Python version: 3.9.15 | packaged by conda-forge | (main, Nov 22 2022, 15:55:03) [GCC 10.4.0] (64-bit runtime)
Python platform: Linux-5.15.0-60-generic-x86_64-with-glibc2.31
Is CUDA available: True
CUDA runtime version: 11.7.99
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration:
GPU 0: NVIDIA RTX A6000
GPU 1: NVIDIA RTX A6000
GPU 2: NVIDIA RTX A6000
Nvidia driver version: 470.161.03
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
Byte Order: Little Endian
Address sizes: 43 bits physical, 48 bits virtual
CPU(s): 48
On-line CPU(s) list: 0-47
Thread(s) per core: 2
Core(s) per socket: 24
Socket(s): 1
NUMA node(s): 1
Vendor ID: AuthenticAMD
CPU family: 23
Model: 49
Model name: AMD Ryzen Threadripper 3960X 24-Core Processor
Stepping: 0
Frequency boost: enabled
CPU MHz: 2200.000
CPU max MHz: 4568.1641
CPU min MHz: 2200.0000
BogoMIPS: 7600.15
Virtualization: AMD-V
L1d cache: 768 KiB
L1i cache: 768 KiB
L2 cache: 12 MiB
L3 cache: 128 MiB
NUMA node0 CPU(s): 0-47
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Retbleed: Mitigation; untrained return thunk; SMT enabled with STIBP protection
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
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
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
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 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd 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 sme sev sev_es
Versions of relevant libraries:
[pip3] ema-pytorch==0.0.10
[pip3] mypy==0.971
[pip3] mypy-extensions==0.4.3
[pip3] numpy==1.23.5
[pip3] pytorch-lightning==1.8.0
[pip3] pytorch-ranger==0.1.1
[pip3] separate-torch==0.0.0
[pip3] torch==1.13.0
[pip3] torch-optimizer==0.3.0
[pip3] torch-summary==1.4.5
[pip3] torchaudio==0.13.0
[pip3] torchmetrics==0.10.1
[pip3] torchvision==0.14.0
[conda] blas 1.0 mkl conda-forge
[conda] cudatoolkit 11.7.0 hd8887f6_10 nvidia
[conda] ema-pytorch 0.0.10 pypi_0 pypi
[conda] libblas 3.9.0 16_linux64_mkl conda-forge
[conda] libcblas 3.9.0 16_linux64_mkl conda-forge
[conda] liblapack 3.9.0 16_linux64_mkl conda-forge
[conda] mkl 2022.1.0 hc2b9512_224
[conda] numpy 1.23.5 py39h3d75532_0 conda-forge
[conda] pytorch 1.13.0 py3.9_cuda11.7_cudnn8.5.0_0 pytorch
[conda] pytorch-cuda 11.7 h67b0de4_1 pytorch
[conda] pytorch-lightning 1.8.0 pypi_0 pypi
[conda] pytorch-mutex 1.0 cuda pytorch
[conda] pytorch-ranger 0.1.1 pyhd8ed1ab_0 conda-forge
[conda] separate-torch 0.0.0 pypi_0 pypi
[conda] torch-optimizer 0.3.0 pyhd8ed1ab_0 conda-forge
[conda] torch-summary 1.4.5 pypi_0 pypi
[conda] torchaudio 0.13.0 py39_cu117 pytorch
[conda] torchmetrics 0.10.1 pyhd8ed1ab_0 conda-forge
[conda] torchvision 0.14.0 py39_cu117 pytorch
cc @ezyang @albanD @zou3519 @gqchen @pearu @nikitaved @soulitzer @lezcano @Varal7 @soumith @msaroufim @wconstab @ngimel @bdhirsh