explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-11-04 01:48:04 +03:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #5518 from svlandeg/fix/pretrain-docs

Browse Source 
Pretrain fixes
This commit is contained in:
Matthew Honnibal 2020-05-29 19:20:20 +02:00 committed by GitHub
commit 2a8137aba9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 22 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
examples/training/pretrain_textcat.py
View File

@ -187,7 +187,7 @@ def evaluate_textcat(tokenizer, textcat, texts, cats):
width=("Width of CNN layers", "positional", None, int), width=("Width of CNN layers", "positional", None, int),
embed_size=("Embedding rows", "positional", None, int), embed_size=("Embedding rows", "positional", None, int),
pretrain_iters=("Number of iterations to pretrain", "option", "pn", int), pretrain_iters=("Number of iterations to pretrain", "option", "pn", int),
train_iters=("Number of iterations to pretrain", "option", "tn", int), train_iters=("Number of iterations to train", "option", "tn", int),
train_examples=("Number of labelled examples", "option", "eg", int), train_examples=("Number of labelled examples", "option", "eg", int),
vectors_model=("Name or path to vectors model to learn from"), vectors_model=("Name or path to vectors model to learn from"),
) )

9
spacy/cli/train.py
View File

@ -15,6 +15,7 @@ import random
from .._ml import create_default_optimizer from .._ml import create_default_optimizer
from ..util import use_gpu as set_gpu from ..util import use_gpu as set_gpu
from ..errors import Errors
from ..gold import GoldCorpus from ..gold import GoldCorpus
from ..compat import path2str from ..compat import path2str
from ..lookups import Lookups from ..lookups import Lookups
@ -182,6 +183,7 @@ def train(
msg.warn("Unable to activate GPU: {}".format(use_gpu)) msg.warn("Unable to activate GPU: {}".format(use_gpu))
msg.text("Using CPU only") msg.text("Using CPU only")
use_gpu = -1 use_gpu = -1
base_components = []
if base_model: if base_model:
msg.text("Starting with base model '{}'".format(base_model)) msg.text("Starting with base model '{}'".format(base_model))
nlp = util.load_model(base_model) nlp = util.load_model(base_model)
@ -227,6 +229,7 @@ def train(
exits=1, exits=1,
) )
msg.text("Extending component from base model '{}'".format(pipe)) msg.text("Extending component from base model '{}'".format(pipe))
base_components.append(pipe)
disabled_pipes = nlp.disable_pipes( disabled_pipes = nlp.disable_pipes(
[p for p in nlp.pipe_names if p not in pipeline] [p for p in nlp.pipe_names if p not in pipeline]
) )
@ -299,7 +302,7 @@ def train(
# Load in pretrained weights # Load in pretrained weights
if init_tok2vec is not None: if init_tok2vec is not None:
components = _load_pretrained_tok2vec(nlp, init_tok2vec) components = _load_pretrained_tok2vec(nlp, init_tok2vec, base_components)
msg.text("Loaded pretrained tok2vec for: {}".format(components)) msg.text("Loaded pretrained tok2vec for: {}".format(components))
# Verify textcat config # Verify textcat config
@ -642,7 +645,7 @@ def _load_vectors(nlp, vectors):
util.load_model(vectors, vocab=nlp.vocab) util.load_model(vectors, vocab=nlp.vocab)
def _load_pretrained_tok2vec(nlp, loc): def _load_pretrained_tok2vec(nlp, loc, base_components):
"""Load pretrained weights for the 'token-to-vector' part of the component """Load pretrained weights for the 'token-to-vector' part of the component
models, which is typically a CNN. See 'spacy pretrain'. Experimental. models, which is typically a CNN. See 'spacy pretrain'. Experimental.
""" """
@ -651,6 +654,8 @@ def _load_pretrained_tok2vec(nlp, loc):
loaded = [] loaded = []
for name, component in nlp.pipeline: for name, component in nlp.pipeline:
if hasattr(component, "model") and hasattr(component.model, "tok2vec"): if hasattr(component, "model") and hasattr(component.model, "tok2vec"):
if name in base_components:
raise ValueError(Errors.E200.format(component=name))
component.tok2vec.from_bytes(weights_data) component.tok2vec.from_bytes(weights_data)
loaded.append(name) loaded.append(name)
return loaded return loaded

2
spacy/errors.py
View File

@ -568,6 +568,8 @@ class Errors(object):
E198 = ("Unable to return {n} most similar vectors for the current vectors " E198 = ("Unable to return {n} most similar vectors for the current vectors "
"table, which contains {n_rows} vectors.") "table, which contains {n_rows} vectors.")
E199 = ("Unable to merge 0-length span at doc[{start}:{end}].") E199 = ("Unable to merge 0-length span at doc[{start}:{end}].")
E200 = ("Specifying a base model with a pretrained component '{component}' "
"can not be combined with adding a pretrained Tok2Vec layer.")
@add_codes @add_codes

24
website/docs/api/cli.md
View File

@ -455,7 +455,7 @@ improvement.
```bash ```bash
$ python -m spacy pretrain [texts_loc] [vectors_model] [output_dir] $ python -m spacy pretrain [texts_loc] [vectors_model] [output_dir]
[--width] [--depth] [--cnn-window] [--cnn-pieces] [--use-chars] [--sa-depth] [--width] [--conv-depth] [--cnn-window] [--cnn-pieces] [--use-chars] [--sa-depth]
[--embed-rows] [--loss_func] [--dropout] [--batch-size] [--max-length] [--embed-rows] [--loss_func] [--dropout] [--batch-size] [--max-length]
[--min-length] [--seed] [--n-iter] [--use-vectors] [--n-save-every] [--min-length] [--seed] [--n-iter] [--use-vectors] [--n-save-every]
[--init-tok2vec] [--epoch-start] [--init-tok2vec] [--epoch-start]
@ -467,7 +467,7 @@ $ python -m spacy pretrain [texts_loc] [vectors_model] [output_dir]
| `vectors_model` | positional | Name or path to spaCy model with vectors to learn from. | | `vectors_model` | positional | Name or path to spaCy model with vectors to learn from. |
| `output_dir` | positional | Directory to write models to on each epoch. | | `output_dir` | positional | Directory to write models to on each epoch. |
| `--width`, `-cw` | option | Width of CNN layers. | | `--width`, `-cw` | option | Width of CNN layers. |
| `--depth`, `-cd` | option | Depth of CNN layers. | | `--conv-depth`, `-cd` | option | Depth of CNN layers. |
| `--cnn-window`, `-cW` <Tag variant="new">2.2.2</Tag> | option | Window size for CNN layers. | | `--cnn-window`, `-cW` <Tag variant="new">2.2.2</Tag> | option | Window size for CNN layers. |
| `--cnn-pieces`, `-cP` <Tag variant="new">2.2.2</Tag> | option | Maxout size for CNN layers. `1` for [Mish](https://github.com/digantamisra98/Mish). | | `--cnn-pieces`, `-cP` <Tag variant="new">2.2.2</Tag> | option | Maxout size for CNN layers. `1` for [Mish](https://github.com/digantamisra98/Mish). |
| `--use-chars`, `-chr` <Tag variant="new">2.2.2</Tag> | flag | Whether to use character-based embedding. | | `--use-chars`, `-chr` <Tag variant="new">2.2.2</Tag> | flag | Whether to use character-based embedding. |
@ -541,16 +541,16 @@ $ python -m spacy init-model [lang] [output_dir] [--jsonl-loc] [--vectors-loc]
[--prune-vectors] [--prune-vectors]
``` ```
| Argument | Type | Description | | Argument | Type | Description |
| ----------------------- | ---------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | -------------------------- | ---------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `lang` | positional | Model language [ISO code](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes), e.g. `en`. | | `lang` | positional | Model language [ISO code](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes), e.g. `en`. |
| `output_dir` | positional | Model output directory. Will be created if it doesn't exist. | | `output_dir` | positional | Model output directory. Will be created if it doesn't exist. |
| `--jsonl-loc`, `-j` | option | Optional location of JSONL-formatted [vocabulary file](/api/annotation#vocab-jsonl) with lexical attributes. | | `--jsonl-loc`, `-j` | option | Optional location of JSONL-formatted [vocabulary file](/api/annotation#vocab-jsonl) with lexical attributes. |
| `--vectors-loc`, `-v` | option | Optional location of vectors. Should be a file where the first row contains the dimensions of the vectors, followed by a space-separated Word2Vec table. File can be provided in `.txt` format or as a zipped text file in `.zip` or `.tar.gz` format. | | `--vectors-loc`, `-v` | option | Optional location of vectors. Should be a file where the first row contains the dimensions of the vectors, followed by a space-separated Word2Vec table. File can be provided in `.txt` format or as a zipped text file in `.zip` or `.tar.gz` format. |
| `--truncate-vectors`, `-t` | option | Number of vectors to truncate to when reading in vectors file. Defaults to `0` for no truncation. | | `--truncate-vectors`, `-t` | option | Number of vectors to truncate to when reading in vectors file. Defaults to `0` for no truncation. |
| `--prune-vectors`, `-V` | option | Number of vectors to prune the vocabulary to. Defaults to `-1` for no pruning. | | `--prune-vectors`, `-V` | option | Number of vectors to prune the vocabulary to. Defaults to `-1` for no pruning. |
| `--vectors-name`, `-vn` | option | Name to assign to the word vectors in the `meta.json`, e.g. `en_core_web_md.vectors`. | | `--vectors-name`, `-vn` | option | Name to assign to the word vectors in the `meta.json`, e.g. `en_core_web_md.vectors`. |
| **CREATES** | model | A spaCy model containing the vocab and vectors. | | **CREATES** | model | A spaCy model containing the vocab and vectors. |
## Evaluate {#evaluate new="2"} ## Evaluate {#evaluate new="2"}