ai-toolkit
ai-toolkit copied to clipboard
Slider text training error. AttributeError: 'int' object has no attribute 'int'
I get this error using default yml config It happens after the Generating baseline samples.
Error running job: 'int' object has no attribute 'int'
========================================
Result:
- 0 completed jobs
- 1 failure
========================================
Traceback (most recent call last):
File "S:\ai-toolkit\run.py", line 89, in <module>
main()
File "S:\ai-toolkit\run.py", line 85, in main
raise e
File "S:\ai-toolkit\run.py", line 77, in main
job.run()
File "S:\ai-toolkit\jobs\TrainJob.py", line 49, in run
process.run()
File "S:\ai-toolkit\jobs\process\BaseSDTrainProcess.py", line 1495, in run
loss_dict = self.hook_train_loop(batch)
File "S:\ai-toolkit\jobs\process\TrainSliderProcess.py", line 653, in hook_train_loop
loss = apply_snr_weight(loss, timesteps_index_list, noise_scheduler,
File "S:\ai-toolkit\toolkit\train_tools.py", line 791, in apply_snr_weight
snr = torch.stack([all_snr[(t - offset).int()] for t in timesteps])
File "S:\ai-toolkit\toolkit\train_tools.py", line 791, in <listcomp>
snr = torch.stack([all_snr[(t - offset).int()] for t in timesteps])
AttributeError: 'int' object has no attribute 'int'
@specblades I had the same error.
I solved it by removing .int() from line 791 of train_tools.py, which was giving the error.
I don't know if it has any other implications though, but it worked for now.
Did this dev die or just give a you know what? I just ran into this error and this is July 20 and your comment is from Jan 28. WTH?