jzhanson
commented
3 years ago FWIW/FYI when running a torch/torchvision checkpoint from 1.6.0/0.7.0 with the torch version 1.1.0/0.3.0 of the ALFRED requirements.txt, it gives the following error
{'tests_seen': 1533,
'tests_unseen': 1529,
'train': 21023,
'valid_seen': 820,
'valid_unseen': 821}
Loading: exp/model:seq2seq_im_mask,name:pm_and_subgoals_01/best_seen.pth
Traceback (most recent call last):
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 181, in nti
s = nts(s, "ascii", "strict")
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 165, in nts
return s.decode(encoding, errors)
UnicodeDecodeError: 'ascii' codec can't decode byte 0xba in position 1: ordinal not in range(128)
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 2281, in next
tarinfo = self.tarinfo.fromtarfile(self)
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 1083, in fromtarfile
obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors)
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 1025, in frombuf
chksum = nti(buf[148:156])
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 184, in nti
raise InvalidHeaderError("invalid header")
tarfile.InvalidHeaderError: invalid header
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/jzhanson/alfred_env/lib/python3.5/site-packages/torch/serialization.py", line 556, in _load
return legacy_load(f)
File "/home/jzhanson/alfred_env/lib/python3.5/site-packages/torch/serialization.py", line 467, in legacy_load
with closing(tarfile.open(fileobj=f, mode='r:', format=tarfile.PAX_FORMAT)) as tar, \
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 1577, in open
return func(name, filemode, fileobj, **kwargs)
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 1607, in taropen
return cls(name, mode, fileobj, **kwargs)
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 1472, in __init__
self.firstmember = self.next()
File "/home/jzhanson/alfred_env/lib/python3.5/tarfile.py", line 2293, in next
raise ReadError(str(e))
tarfile.ReadError: invalid header
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "models/eval/eval_seq2seq.py", line 54, in <module>
eval = EvalTask(args, manager)
File "/home/jzhanson/alfred/models/eval/eval.py", line 31, in __init__
self.model, optimizer = M.Module.load(self.args.model_path)
File "/home/jzhanson/alfred/models/model/seq2seq.py", line 318, in load
save = torch.load(fsave)
File "/home/jzhanson/alfred_env/lib/python3.5/site-packages/torch/serialization.py", line 387, in load
return _load(f, map_location, pickle_module, **pickle_load_args)
File "/home/jzhanson/alfred_env/lib/python3.5/site-packages/torch/serialization.py", line 560, in _load
raise RuntimeError("{} is a zip archive (did you mean to use torch.jit.load()?)".format(f.name))
RuntimeError: exp/model:seq2seq_im_mask,name:pm_and_subgoals_01/best_seen.pth is a zip archive (did you mean to use torch.jit.load()?)
MohitShridhar
When running evaluation with the pretrained models (or training) inside the ALFRED docker (
$ python3 models/eval/eval_seq2seq.py --model_path exp/model:seq2seq_im_mask,name:base30_pm010_sg010_01/best_seen.pth --eval_split valid_seen --data data/json_feat_2.1.0 --model models.model.seq2seq_im_mask --gpu --num_threads 1) there seems to be a torch/torchvision compatibility problem with drivers (which may be due to a wonky driver setup on my end). If I'm running the pretrained model inside ai2thor-docker instead, I get the same error but I can update torch and torchvision to 1.6.0 and 0.7.0 respectively and the error goes away, leading to this issue where the Unity process crashes immediately due to a driver mismatch.Currently having a bit of difficulty building the ALFRED docker with a newer python version (>= 3.6) that would allow me to upgrade torch and torchvision, but that's more of a "me" problem.