AssertionError: only one Trainer can be instantiated at a time for training

[tiansiyuan]

With version 1.2.16, when I run audiolm_pytorch_demo.ipynb, after I successfully complete training SoundStream and start training SemanticTransformer, I get the error message as shown in title.

The full error message is as below.

Thanks,

Tian

---

AssertionError Traceback (most recent call last)
Cell In[5], line 24
12 wav2vec = HubertWithKmeans(
13 checkpoint_path = f'./{hubert_ckpt}',
14 kmeans_path = f'./{hubert_quantizer}'
15 )
17 semantic_transformer = SemanticTransformer(
18 num_semantic_tokens = wav2vec.codebook_size,
19 dim = 1024,
20 depth = 6
21 )# .cuda()
---> 24 trainer = SemanticTransformerTrainer(
25 transformer = semantic_transformer,
26 wav2vec = wav2vec,
27 folder = dataset_folder,
28 batch_size = 1,
29 data_max_length = 320 * 32,
30 num_train_steps = 1
31 )
33 trainer.train()

File <@beartype(audiolm_pytorch.trainer.SemanticTransformerTrainer.init) at 0x7f2258c97670>:111, in init(__beartype_func, __beartype_conf, __beartype_get_violation, __beartype_object_139785495343936, __beartype_object_205596576, __beartype_object_139785495313344, __beartype_object_9748320, __beartype_object_115133280, *args, **kwargs)

File /home/venv/lib/python3.9/site-packages/audiolm_pytorch/trainer.py:579, in SemanticTransformerTrainer.init(self, wav2vec, transformer, num_train_steps, batch_size, audio_conditioner, dataset, data_max_length, data_max_length_seconds, folder, lr, grad_accum_every, wd, max_grad_norm, valid_frac, random_split_seed, save_results_every, save_model_every, results_folder, accelerate_kwargs, force_clear_prev_results)
553 @beartype
554 def init(
555 self,
(...)
576 force_clear_prev_results = None
577 ):
578 super().init()
--> 579 check_one_trainer()
581 self.accelerator = Accelerator(**accelerate_kwargs)
583 self.wav2vec = wav2vec

File /home/venv/lib/python3.9/site-packages/audiolm_pytorch/trainer.py:59, in check_one_trainer()
57 def check_one_trainer():
58 global ONE_TRAINER_INSTANTIATED
---> 59 assert not ONE_TRAINER_INSTANTIATED, 'only one Trainer can be instantiated at a time for training'
60 ONE_TRAINER_INSTANTIATED = True

AssertionError: only one Trainer can be instantiated at a time for training

[LWprogramming]

#211