Closed heury closed 1 year ago
ClapRVQTrainer does not have "log_with" and it causes error in following line
if 'tensorboard' in self.log_with: self.accelerator.init_trackers(f"claprvq{int(time.time() * 1000)}", config=hps) else: self.accelerator.init_trackers(f"clap_rvq", config=hps)
So, I added log_with refer to SingleStageTrainer class and it works fine.
self.log_with = accelerate_kwargs['log_with'] if 'log_with' in accelerate_kwargs else None
my bad! fixed in 0e3ada6
ClapRVQTrainer does not have "log_with" and it causes error in following line
if 'tensorboard' in self.log_with: self.accelerator.init_trackers(f"claprvq{int(time.time() * 1000)}", config=hps) else: self.accelerator.init_trackers(f"clap_rvq", config=hps)
So, I added log_with refer to SingleStageTrainer class and it works fine.
self.log_with = accelerate_kwargs['log_with'] if 'log_with' in accelerate_kwargs else None