facebookresearch / GuidedDistillation

Official implementation of the paper "Guided Distillation for Semi-Supervised Instance Segmentation".
Other
35 stars 4 forks source link

SSL Teacher training error #1

Closed anirudh6415 closed 10 months ago

anirudh6415 commented 10 months ago

can you please help me resolve this issue @TariqBerrada

Traceback (most recent call last): File "..\GuidedDistillation\train_net.py", line 430, in launch( File "\detectron2\detectron2\engine\launch.py", line 84, in the launch main_func(*args) File "\GuidedDistillation\train_net.py", line 424, in main return trainer.train() File "\GuidedDistillation\modules\defaults.py", line 570, in train super().train(self.start_iter, self.max_iter) File "\GuidedDistillation\modules\train_loop.py", line 164, in train self.run_step() File "\GuidedDistillation\modules\defaults.py", line 580, in run_step self._trainer.run_step() File "\GuidedDistillation\modules\train_loop.py", line 901, in run_step if self.model.module.do_ssl and self.model.module.iter % self.model.module.ssl_freq == 0 : File ".conda\envs\nlp\lib\site-packages\torch\nn\modules\module.py", line 1695, in getattr raise AttributeError(f"'{type(self).name}' object has no attribute '{name}'") AttributeError: 'MaskFormer' object has no attribute 'module'

TariqBerrada commented 10 months ago

Hello, This is probably due to launching the job on one gpu only, which is not yet supported on the codebase. If you try to launch the job on two or more gpus then it should not throw this error.