AttributeError: 'HashEmbed' object has no attribute 'G'

[damianoporta]

Hello, i am training a new NER model with the code:

with nlp.disable_pipes(*other_pipes):  # only train NER
    # reset and initialize the weights randomly – but only if we're
    # training a new model
    nlp.begin_training(component_cfg={"ner": {"conv_depth": 8}})

    for itn in range(N_ITER):
        random.shuffle(TRAIN_DATA)
        losses = {}
        # batch up the examples using spaCy's minibatch
        batches = minibatch(TRAIN_DATA, size=compounding(4.0, 32.0, 1.001))

        for batch in batches:
            texts, annotations = zip(*batch)
            nlp.update(
                texts,  # batch of texts
                annotations,  # batch of annotations
                drop=0.2,  # dropout - make it harder to memorise data
                losses=losses,
            )

i only have changed the conv_depth setting as said by @honnibal here: https://github.com/explosion/spaCy/issues/3798

Then, loading the model i get:

>>> nlp = spacy.load('/home/nlp/0')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy/__init__.py", line 27, in load
    return util.load_model(name, **overrides)
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy/util.py", line 133, in load_model
    return load_model_from_path(Path(name), **overrides)
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy/util.py", line 173, in load_model_from_path
    return nlp.from_disk(model_path)
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy/language.py", line 791, in from_disk
    util.from_disk(path, deserializers, exclude)
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy/util.py", line 630, in from_disk
    reader(path / key)
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy/language.py", line 787, in <lambda>
    deserializers[name] = lambda p, proc=proc: proc.from_disk(p, exclude=["vocab"])
  File "nn_parser.pyx", line 634, in spacy.syntax.nn_parser.Parser.from_disk
  File "/home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/thinc/neural/_classes/model.py", line 371, in from_bytes
    dest = getattr(layer, name)
AttributeError: 'HashEmbed' object has no attribute 'G'

What is this problem?

Your Environment

(.prodigy-1.8.3) root@damiano:/home/nlp# python -m spacy info

============================== Info about spaCy ==============================

spaCy version    2.1.4                         
Location         /home/nlp/.prodigy-1.8.3/lib/python3.6/site-packages/spacy
Platform         Linux-4.15.0-55-generic-x86_64-with-Ubuntu-18.04-bionic
Python version   3.6.7                         
Models

[ines]

Did you also add the {"conv_depth": 8} setting to your ner component? And when you look at the cfg within /home/nlp/0/ner, is it included there?

[damianoporta]

No, do I need to add it manually? I only have added the setting in begin_training

[alejandrojcastaneira]

@damianoporta could you try settingos.environ["conv_depth"]="8" before loading the model?

[damianoporta]

@alejandrojcastaneira it works!

os.environ['conv_depth'] = "8" nlp = spacy.load('/home/nlp/0')

Now the question is....why?

@ines in ner/cfg there is no conv_depth setting. this is what i see:

{
  "beam_width":1,
  "beam_density":0.0,
  "beam_update_prob":1.0,
  "cnn_maxout_pieces":3,
  "nr_class":89,
  "hidden_depth":1,
  "token_vector_width":96,
  "hidden_width":64,
  "maxout_pieces":2,
  "pretrained_vectors":"it_model.vectors",
  "bilstm_depth":0
}

why the setting is not saved when i do nlp.to_disk() ?

[damianoporta]

Did you also add the {"conv_depth": 8} setting to your ner component? And when you look at the cfg within /home/nlp/0/ner, is it included there?

@ines i have added: that parameter on ner/cfg and it works. Do i need it somewhere else?

[ines]

that parameter on ner/cfg and it works. Do i need it somewhere else?

Sorry, I hadn't seen your first reply. Glad it works!

You should only have to add it to the component so it can be serialized/deserialized correctly. You don't have to hack it into the cfg manually, though – you should also be able to set it when you create the ner component:

ner = nlp.create_pipe("ner", config={"conv_depth": 8})

[lock[bot]]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.