Running Custom NER training using local en_core_web_lg in config file

[iteimouri]

I've made the following config file to make a custom NER:

[paths]
train = null
dev = null
vectors = null
init_tok2vec = null

[system]
gpu_allocator = null
seed = 0

[nlp]
lang = "en"
pipeline = ["tok2vec","ner"]
batch_size = 1000
disabled = []
before_creation = null
after_creation = null
after_pipeline_creation = null
tokenizer = {"@tokenizers":"spacy.Tokenizer.v1"}

[components]

[components.ner]
factory = "ner"
incorrect_spans_key = null
moves = null
scorer = {"@scorers":"spacy.ner_scorer.v1"}
update_with_oracle_cut_size = 100

[components.ner.model]
@architectures = "spacy.TransitionBasedParser.v2"
state_type = "ner"
extra_state_tokens = false
hidden_width = 64
maxout_pieces = 2
use_upper = true
nO = null

[components.ner.model.tok2vec]
@architectures = "spacy.Tok2VecListener.v1"
width = ${components.tok2vec.model.encode.width}
upstream = "*"

[components.tok2vec]
factory = "tok2vec"

[components.tok2vec.model]
@architectures = "spacy.Tok2Vec.v2"

[components.tok2vec.model.embed]
@architectures = "spacy.MultiHashEmbed.v2"
width = ${components.tok2vec.model.encode.width}
attrs = ["NORM","PREFIX","SUFFIX","SHAPE"]
rows = [5000,2500,2500,2500]
include_static_vectors = true

[components.tok2vec.model.encode]
@architectures = "spacy.MaxoutWindowEncoder.v2"
width = 256
depth = 8
window_size = 1
maxout_pieces = 3

[corpora]

[corpora.dev]
@readers = "spacy.Corpus.v1"
path = ${paths.dev}
max_length = 0
gold_preproc = false
limit = 0
augmenter = null

[corpora.train]
@readers = "spacy.Corpus.v1"
path = ${paths.train}
max_length = 0
gold_preproc = false
limit = 0
augmenter = null

[training]
dev_corpus = "corpora.dev"
train_corpus = "corpora.train"
seed = ${system.seed}
gpu_allocator = ${system.gpu_allocator}
dropout = 0.1
accumulate_gradient = 1
patience = 1600
max_epochs = 0
max_steps = 20000
eval_frequency = 200
frozen_components = []
annotating_components = []
before_to_disk = null

[training.batcher]
@batchers = "spacy.batch_by_words.v1"
discard_oversize = false
tolerance = 0.2
get_length = null

[training.batcher.size]
@schedules = "compounding.v1"
start = 100
stop = 1000
compound = 1.001
t = 0.0

[training.logger]
@loggers = "spacy.ConsoleLogger.v1"
progress_bar = false

[training.optimizer]
@optimizers = "Adam.v1"
beta1 = 0.9
beta2 = 0.999
L2_is_weight_decay = true
L2 = 0.01
grad_clip = 1.0
use_averages = false
eps = 0.00000001
learn_rate = 0.001

[training.score_weights]
ents_f = 1.0
ents_p = 0.0
ents_r = 0.0
ents_per_type = null

[pretraining]

[initialize]
vectors = "P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\vocab\vectors"
init_tok2vec = ${paths.init_tok2vec}
vocab_data = null
lookups = null
before_init = null
after_init = null

[initialize.components]

[initialize.tokenizer]

Towards the end I had to change the vectors path from en_core_web_lg to "P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\vocab\vectors" this is because I can not fetch the en_core_web_lg model from online servers but only can do that locally (due to work restriction). However I get the following error:

[2022-01-27 16:21:16,595] [INFO] Set up nlp object from config
[2022-01-27 16:21:16,608] [INFO] Pipeline: ['tok2vec', 'ner']
[2022-01-27 16:21:16,613] [INFO] Created vocabulary
Traceback (most recent call last):
  File "P:\myPythonEnv\lib\site-packages\spacy\language.py", line 1283, in initialize
    init_vocab(
  File "P:\myPythonEnv\lib\site-packages\spacy\training\initialize.py", line 131, in init_vocab
    load_vectors_into_model(nlp, vectors)
  File "P:\myPythonEnv\lib\site-packages\spacy\training\initialize.py", line 152, in load_vectors_into_model
    vectors_nlp = load_model(name, vocab=nlp.vocab, exclude=exclude)
  File "P:\myPythonEnv\lib\site-packages\spacy\util.py", line 427, in load_model
    raise IOError(Errors.E050.format(name=name))
OSError: [E050] Can't find model '"P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\vocab\vectors"'. It doesn't seem to be a Python package or a valid path to a data directory.

During handling of the above exception, another exception occurred:
Traceback (most recent call last):
  File "P:\myPythonEnv\lib\runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "P:\myPythonEnv\lib\runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "P:\myPythonEnv\lib\site-packages\spacy\__main__.py", line 4, in <module>
    setup_cli()
  File "P:\myPythonEnv\lib\site-packages\spacy\cli\_util.py", line 71, in setup_cli
    command(prog_name=COMMAND)
  File "P:\myPythonEnv\lib\site-packages\click\core.py", line 829, in __call__
    return self.main(*args, **kwargs)
  File "P:\myPythonEnv\lib\site-packages\click\core.py", line 782, in main
    rv = self.invoke(ctx)
  File "P:\myPythonEnv\lib\site-packages\click\core.py", line 1259, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "P:\myPythonEnv\lib\site-packages\click\core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "P:\myPythonEnv\lib\site-packages\click\core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "P:\myPythonEnv\lib\site-packages\typer\main.py", line 497, in wrapper
    return callback(**use_params)  # type: ignore
  File "P:\myPythonEnv\lib\site-packages\spacy\cli\train.py", line 45, in train_cli
    train(config_path, output_path, use_gpu=use_gpu, overrides=overrides)
  File "P:\myPythonEnv\lib\site-packages\spacy\cli\train.py", line 72, in train
    nlp = init_nlp(config, use_gpu=use_gpu)
  File "P:\myPythonEnv\lib\site-packages\spacy\training\initialize.py", line 84, in init_nlp
    nlp.initialize(lambda: train_corpus(nlp), sgd=optimizer)
  File "P:\myPythonEnv\lib\site-packages\spacy\language.py", line 1287, in initialize
    raise IOError(Errors.E884.format(vectors=I["vectors"]))
OSError: [E884] The pipeline could not be initialized because the vectors could not be found at '"P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\vocab\vectors"'. If your pipeline was already initialized/trained before, call 'resume_training' instead of 'initialize', or initialize only the components that are new.

I have tried various paths and all of the gave the same error - these non-working paths are:

"P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\vocab"

"P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\vocab\vectors"

"P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0"

"en_core_web_lg"

I wonder if there is a way to fetch the model using some proxy in the config file or how can I make sure the vectors are loaded locally in a correct manner?

My Spacy version is 3.2.1.

[pmbaumgartner]

Hey @iteimouri - I'm going to move this to a discussion since I don't believe it is a bug.

I think you need to specify the model directory two levels up from the folder you currently have for the vectors. So the relevant config line should read:

vectors = "P:\Documents\Python Scripts\newPythonEnv\Additive_NER_Spacy3\en_core_web_lg-3.2.0.tar\en_core_web_lg-3.2.0\en_core_web_lg\en_core_web_lg-3.2.0\"

[iteimouri]

sure :)

[pmbaumgartner]

Sorry - I just edited my response to include the answer, I didn't realize GitHub would make my comment on the Issue a top-level reply.

[pmbaumgartner]

I see that you've tried out additional paths as well. I'm now wondering if it's some sort of Windows path issues, because I replicated this exact situation and used a full path to that folder and things worked as I expected.

Here is my local file structure:

.
├── config.cfg
├── en_core_web_lg-3.2.0
│  ├── en_core_web_lg
│  │  ├── __init__.py
│  │  ├── en_core_web_lg-3.2.0
│  │  │  ├── accuracy.json
│  │  │  ├── attribute_ruler
│  │  │  │  └── patterns
│  │  │  ├── config.cfg
│  │  │  ├── lemmatizer
│  │  │  │  └── lookups
│  │  │  │     └── lookups.bin
│  │  │  ├── LICENSE
│  │  │  ├── LICENSES_SOURCES
│  │  │  ├── meta.json
│  │  │  ├── ner
│  │  │  │  ├── cfg
│  │  │  │  ├── model
│  │  │  │  └── moves
│  │  │  ├── parser
│  │  │  │  ├── cfg
│  │  │  │  ├── model
│  │  │  │  └── moves
│  │  │  ├── README.md
│  │  │  ├── senter
│  │  │  │  ├── cfg
│  │  │  │  └── model
│  │  │  ├── tagger
│  │  │  │  ├── cfg
│  │  │  │  └── model
│  │  │  ├── tok2vec
│  │  │  │  ├── cfg
│  │  │  │  └── model
│  │  │  ├── tokenizer
│  │  │  └── vocab
│  │  │     ├── key2row
│  │  │     ├── lookups.bin
│  │  │     ├── strings.json
│  │  │     ├── vectors
│  │  │     └── vectors.cfg
│  │  └── meta.json
│  ├── en_core_web_lg.egg-info
│  │  ├── dependency_links.txt
│  │  ├── entry_points.txt
│  │  ├── not-zip-safe
│  │  ├── PKG-INFO
│  │  ├── requires.txt
│  │  ├── SOURCES.txt
│  │  └── top_level.txt
│  ├── LICENSE
│  ├── LICENSES_SOURCES
│  ├── MANIFEST.in
│  ├── meta.json
│  ├── PKG-INFO
│  ├── README.md
│  ├── setup.cfg
│  └── setup.py
└── en_core_web_lg-3.2.0.tar.gz

And running the train command gives me the following:

 python -m spacy train config.cfg --paths.train ./train.spacy --paths.dev ./dev.spacy
ℹ No output directory provided
ℹ Using CPU

=========================== Initializing pipeline ===========================
[2022-01-27 15:44:16,427] [INFO] Set up nlp object from config
[2022-01-27 15:44:16,434] [INFO] Pipeline: ['tok2vec', 'ner']
[2022-01-27 15:44:16,437] [INFO] Created vocabulary
[2022-01-27 15:44:17,932] [INFO] Added vectors: /Users/peter/projects/issue.spacy.10148/en_core_web_lg-3.2.0/en_core_web_lg/en_core_web_lg-3.2.0

[iteimouri]

Thank you @pmbaumgartner - I've realised my mistake by looking at your path. There was a typo on mine. Now solved!