Problem loading the data.vocab.pt

[SilvesterYu]

Hi! I'm having trouble loading the data checkpoints.

The specific output and error are:

WebNLG
Pre-process training data
Training Set
corpus Read Corpus 0:00:00.000065
graphify RDF to Graph 0:00:00.000024
spelling Fix Spelling 0:00:00.000018
entities Describe entities 0:00:00.000025
match-ents Match Entities 0:00:00.000042
match-plans Match Plans 0:00:00.000020
tokenize Tokenize Plans & Sentences 0:00:00.000017
to-json Export in a readable format 0:00:00.000022
Dev Set
corpus Read Corpus 0:00:00.000042
graphify RDF to Graph 0:00:00.000019
spelling Fix Spelling 0:00:00.000019
entities Describe entities 0:00:00.000020
match-ents Match Entities 0:00:00.000020
match-plans Match Plans 0:00:00.000019
tokenize Tokenize Plans & Sentences 0:00:00.000018
to-json Export in a readable format 0:00:00.000016
Train Planner
planner Learn planner 0:00:00.000036
Train Model
model Initialize OpenNMT 0:00:00.000034
expose Expose Train Data 0:00:00.000017
pre-process Pre-process Train and Dev 0:00:00.000018
train Train model EXEC /home/ubuntu/miniconda3/envs/env_pytorch/bin/python /home/ubuntu/chimera/model/../libs/OpenNMT/train.py -train_steps 30000 -save_checkpoint_steps 1000 -batch_size 16 -word_vec_size 300 -feat_vec_size 10 -feat_merge concat -layers 3 -copy_attn -position_encoding -data /tmp/tmpz79gzk1g/data -save_model /tmp/tmpo6y8o6dz/ -world_size 1 -gpu_ranks 0

------------------------------------------------ (so the error happened here) -------------------------------------------------------

Traceback (most recent call last):
File "/home/ubuntu/chimera/model/../libs/OpenNMT/train.py", line 109, in
main(opt)
File "/home/ubuntu/chimera/model/../libs/OpenNMT/train.py", line 39, in main
single_main(opt, 0)
File "/home/ubuntu/chimera/libs/OpenNMT/onmt/train_single.py", line 63, in main
vocab = torch.load(opt.data + '.vocab.pt')
File "/home/ubuntu/miniconda3/envs/env_pytorch/lib/python3.6/site-packages/torch/serialization.py", line 367, in load
return _load(f, map_location, pickle_module)
File "/home/ubuntu/miniconda3/envs/env_pytorch/lib/python3.6/site-packages/torch/serialization.py", line 528, in _load
magic_number = pickle_module.load(f)
_pickle.UnpicklingError: A load persistent id instruction was encountered,
but no persistent_load function was specified.

--------------------------------------- (the error is above)-------------------------------------

Could you kindly provide some help? I would be very grateful!

[AmitMY]

did it successfully create a vocabulary file in /tmp/tmpz79gzk1g/data?

If there is no persistent tmp storage on your sever, perhaps you should export TMPDIR to some other directory.

[SilvesterYu]

Yes, the vocabulary file is created in /tmp. The files are below:

Also, inside the data.vocab.pt: