Fixing Errors: RuntimeError and ModuleNotFoundError

[doammii]

I referred to the link below to fine-tune the Tortoise model with GPT-2.
During training, I encountered two errors and found ways to fix them.

• Fine-tuning TorToiSe's GPT Model with Colab

Error 1) ModuleNotFoundError: No module named 'torch._six'

Traceback (most recent call last):
File "C:\Users\oldgu\DL-Art-School\codes\train.py", line 12, in
from data.data_sampler import DistIterSampler
File "C:\Users\oldgu\DL-Art-School\codes\data_init_.py", line 6, in
from utils.util import opt_get
File "C:\Users\oldgu\DL-Art-School\codes\utils\util.py", line 25, in
from torch._six import inf
ModuleNotFoundError: No module named 'torch._six'

Solution
: In line 25 of /content/DL-Art-School/codes/utils/util.py

• from torch._six import inf => from torch import inf

Error 2) RuntimeError: Error(s) in loading state_dict for UnifiedVoice

Traceback (most recent call last):
  File "/content/DL-Art-School/codes/train.py", line 398, in <module>
    trainer.init(args.opt, opt, args.launcher)
  File "/content/DL-Art-School/codes/train.py", line 146, in init
    self.model = ExtensibleTrainer(opt)
  File "/content/DL-Art-School/codes/trainer/ExtensibleTrainer.py", line 192, in __init__
    self.load()  # load networks from save states as needed
  File "/content/DL-Art-School/codes/trainer/ExtensibleTrainer.py", line 539, in load
    self.load_network(load_path, net, self.opt['path']['strict_load'], opt_get(self.opt, ['path', f'pretrain_base_path_{name}']))
  File "/content/DL-Art-School/codes/trainer/base_model.py", line 131, in load_network
    network.load_state_dict(load_net_clean, strict=strict)
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 2152, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for UnifiedVoice:
	Unexpected key(s) in state_dict: "gpt.h.0.attn.bias", "gpt.h.0.attn.masked_bias", "gpt.h.1.attn.bias", "gpt.h.1.attn.masked_bias"

Solution
: In line 131 of /content/DL-Art-School/codes/trainer/base_model.py

• network.load_state_dict(load_net_clean, strict=strict) => network.load_state_dict(load_net_clean, strict=False)