TypeError: int() argument must be a string, a bytes-like object or a number, not 'list'

[zakinp]

i got this problem when using "diffusers_trainer.py":

Exception caught on rank 0 at step 10, saving checkpoint...
int() argument must be a string, a bytes-like object or a number, not 'list'
Traceback (most recent call last):
File "/root/waifu-diffusion/trainer/diffusers_trainer.py", line 953, in main
prompt = tokenizer.decode(batch['tokens'][random.randint(0, len(batch['tokens'])-1)])
File "/root/miniconda3/lib/python3.8/site-packages/transformers/tokenization_utils_base.py", line 3367, in decode
return self._decode(
File "/root/miniconda3/lib/python3.8/site-packages/transformers/tokenization_utils.py", line 931, in _decode
filtered_tokens = self.convert_ids_to_tokens(token_ids, skip_special_tokens=skip_special_tokens)
File "/root/miniconda3/lib/python3.8/site-packages/transformers/tokenization_utils.py", line 906, in convert_ids_to_tokens
index = int(index)
TypeError: int() argument must be a string, a bytes-like object or a number, not 'list'

[thojmr]

I got this too when I tried a few days ago, for now you can replace that line with a static prompt until it gets fixed

prompt = "put your prompt here" #tokenizer.decode(batch['tokens'][random.randint(0, len(batch['tokens'])-1)])

[zakinp]

I got this too when I tried a few days ago, for now you can replace that line with a static prompt until it gets fixed

prompt = "put your prompt here" #tokenizer.decode(batch['tokens'][random.randint(0, len(batch['tokens'])-1)])

thanks, i found that set the '--extended_mode_chunks' parameter can also fix this error.

[mcrisafu]

@zakinp To what value did you set it? And what does it? I solved it with setting [0] at the end but this does not seem right...

Thank you and regards

[SY573M404]

@mcrisafu It is a prompt for generating log images from

[mcrisafu]

@SY573M404 Thank you. I meant the --extended_mode_chunks parameter. The question was somewhat misleading. Sry.

[SY573M404]

@mcrisafu It seems that anything greater that 1 would work. It is probably caused by a simple typo here:

waifu-diffusion/trainer/diffusers_trainer.py

Lines 533 to 539 in 1a83e47

	if args.extended_mode_chunks < 2:
	max_length = self.tokenizer.model_max_length - 2
	input_ids = [self.tokenizer([example['input_ids']], truncation=True, return_length=True, return_overflowing_tokens=False, padding=False, add_special_tokens=False, max_length=max_length).input_ids for example in examples if example is not None]
	else:
	max_length = self.tokenizer.model_max_length
	max_chunks = args.extended_mode_chunks
	input_ids = [self.tokenizer([example['input_ids']], truncation=True, return_length=True, return_overflowing_tokens=False, padding=False, add_special_tokens=False, max_length=(max_length * max_chunks) - (max_chunks * 2)).input_ids[0] for example in examples if example is not None]

Like you mentioned before, there is [0] at the end of the second of mostly similar if branches, but not the first. I'll try it and open a PR if it works