pytorch/tnt

AttributeError: module 'torch.amp' has no attribute 'GradScaler'

Opened this issue · 4 comments

djl70 commented

🐛 Describe the bug

Hello, I'm using pytorch 2.2.1 and torchtnt 0.2.3, and the change from #697 seems to be causing an AttributeError for me when trying to import fit.

from torchtnt.framework import fit

Full traceback:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/djl70/miniconda3/envs/tnt-pull-697/lib/python3.10/site-packages/torchtnt/framework/__init__.py", line 7, in <module>
    from .auto_unit import AutoPredictUnit, AutoUnit
  File "/home/djl70/miniconda3/envs/tnt-pull-697/lib/python3.10/site-packages/torchtnt/framework/auto_unit.py", line 19, in <module>
    from torchtnt.framework._loop_utils import _step_requires_iterator
  File "/home/djl70/miniconda3/envs/tnt-pull-697/lib/python3.10/site-packages/torchtnt/framework/_loop_utils.py", line 15, in <module>
    from torchtnt.framework.state import State
  File "/home/djl70/miniconda3/envs/tnt-pull-697/lib/python3.10/site-packages/torchtnt/framework/state.py", line 13, in <module>
    from torchtnt.utils.timer import BoundedTimer, TimerProtocol
  File "/home/djl70/miniconda3/envs/tnt-pull-697/lib/python3.10/site-packages/torchtnt/utils/__init__.py", line 50, in <module>
    from .precision import convert_precision_str_to_dtype
  File "/home/djl70/miniconda3/envs/tnt-pull-697/lib/python3.10/site-packages/torchtnt/utils/precision.py", line 41, in <module>
    ) -> Optional[torch.amp.GradScaler]:
AttributeError: module 'torch.amp' has no attribute 'GradScaler'

Steps to reproduce:

$ conda create -n tnt-pull-697 python=3.10
$ conda activate tnt-pull-697
$ conda install pytorch==2.2.1 torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia
$ conda install -c conda-forge torchtnt==0.2.3
$ python
>>> from torchtnt.framework import fit

Versions

Collecting environment information...
PyTorch version: 2.2.1
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 20.04.6 LTS (x86_64)
GCC version: (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0
Clang version: Could not collect
CMake version: version 3.16.3
Libc version: glibc-2.31

Python version: 3.10.13 (main, Sep 11 2023, 13:44:35) [GCC 11.2.0] (64-bit runtime)
Python platform: Linux-5.15.0-91-generic-x86_64-with-glibc2.31
Is CUDA available: True
CUDA runtime version: 11.1.105
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration:
GPU 0: NVIDIA RTX A5000
GPU 1: NVIDIA RTX A5000
GPU 2: NVIDIA RTX A5000
GPU 3: NVIDIA RTX A5000

Nvidia driver version: 535.129.03
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): 64
On-line CPU(s) list: 0-63
Thread(s) per core: 2
Core(s) per socket: 32
Socket(s): 1
NUMA node(s): 1
Vendor ID: AuthenticAMD
CPU family: 23
Model: 49
Model name: AMD Ryzen Threadripper PRO 3975WX 32-Cores
Stepping: 0
Frequency boost: enabled
CPU MHz: 2200.000
CPU max MHz: 4368.1641
CPU min MHz: 2200.0000
BogoMIPS: 6986.99
Virtualization: AMD-V
L1d cache: 1 MiB
L1i cache: 1 MiB
L2 cache: 16 MiB
L3 cache: 128 MiB
NUMA node0 CPU(s): 0-63
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 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 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 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] mypy-extensions==1.0.0
[pip3] numpy==1.26.4
[pip3] torch==2.2.1
[pip3] torchaudio==2.2.1
[pip3] torchsnapshot==0.1.0
[pip3] torchtnt==0.2.3
[pip3] torchvision==0.17.1
[pip3] triton==2.2.0
[conda] blas 1.0 mkl
[conda] ffmpeg 4.3 hf484d3e_0 pytorch
[conda] libjpeg-turbo 2.0.0 h9bf148f_0 pytorch
[conda] mkl 2023.1.0 h213fc3f_46344
[conda] mkl-service 2.4.0 py310h5eee18b_1
[conda] mkl_fft 1.3.8 py310h5eee18b_0
[conda] mkl_random 1.2.4 py310hdb19cb5_0
[conda] numpy 1.26.4 py310h5f9d8c6_0
[conda] numpy-base 1.26.4 py310hb5e798b_0
[conda] pytorch 2.2.1 py3.10_cuda12.1_cudnn8.9.2_0 pytorch
[conda] pytorch-cuda 12.1 ha16c6d3_5 pytorch
[conda] pytorch-mutex 1.0 cuda pytorch
[conda] torchaudio 2.2.1 py310_cu121 pytorch
[conda] torchsnapshot 0.1.0 pyhd8ed1ab_0 conda-forge
[conda] torchtnt 0.2.3 pyhd8ed1ab_0 conda-forge
[conda] torchtriton 2.2.0 py310 pytorch
[conda] torchvision 0.17.1 py310_cu121 pytorch

galrotem commented

Hey @djl70, thanks for reporting and apologies for this. We'll work on a fix and release a new version.

In the meantime, if you just need fit and you aren't using AutoUnit and you are using just Unit, you can import by "from torchtnt.framework.fit import fit". When you import from torchtnt.framework you get the all of the deps in torchtnt/framework/__init__.py which also contain the AutoUnit.

  • 👍1
FilipinoGambino commented

Hi @galrotem, it looks like the correct import is already at the top of precision.py, so someone just needs to delete "torch.amp." on line 42 and it's fixed!

yiminglin-ai commented

Hi @galrotem, I believe you fixed it in #753, but I don't see a new release. When can we expect version tnt 0.2.4 to be released? Thanks

  • 👍1
JKSenthil commented

Hi @yiminglin-ai, we aim to release within the next day or two, sorry for the delay!

  • 🎉1