Allow overriding meta from spacy.blank

spacy/__init__.py

@@ -47,13 +47,17 @@ def load(
 
 
 def blank(
-    name: str, *, config: Union[Dict[str, Any], Config] = util.SimpleFrozenDict()
+    name: str,
+    *,
+    config: Union[Dict[str, Any], Config] = util.SimpleFrozenDict(),
+    meta: Dict[str, Any] = util.SimpleFrozenDict()
 ) -> Language:
     """Create a blank nlp object for a given language code.
 
     name (str): The language code, e.g. "en".
     config (Dict[str, Any] / Config): Optional config overrides.
+    meta (Dict[str, Any]): Overrides for nlp.meta.
     RETURNS (Language): The nlp object.
     """
     LangClass = util.get_lang_class(name)
-    return LangClass.from_config(config)
+    return LangClass.from_config(config, meta=meta)

spacy/language.py

@@ -1458,6 +1458,7 @@ class Language:
         vocab: Union[Vocab, bool] = True,
         disable: Iterable[str] = SimpleFrozenList(),
         exclude: Iterable[str] = SimpleFrozenList(),
+        meta: Dict[str, Any] = SimpleFrozenDict(),
         auto_fill: bool = True,
         validate: bool = True,
     ) -> "Language":
@@ -1472,6 +1473,7 @@ class Language:
             explicitly enable them by calling nlp.enable_pipe.
         exclude (Iterable[str]): Names of pipeline components to exclude.
             Excluded components won't be loaded.
+        meta (Dict[str, Any]): Meta overrides for nlp.meta.
         auto_fill (bool): Automatically fill in missing values in config based
             on defaults and function argument annotations.
         validate (bool): Validate the component config and arguments against
@@ -1525,7 +1527,7 @@ class Language:
         # inside stuff like the spacy train function. If we loaded them here,
         # then we would load them twice at runtime: once when we make from config,
         # and then again when we load from disk.
-        nlp = lang_cls(vocab=vocab, create_tokenizer=create_tokenizer)
+        nlp = lang_cls(vocab=vocab, create_tokenizer=create_tokenizer, meta=meta)
         if after_creation is not None:
             nlp = after_creation(nlp)
             if not isinstance(nlp, cls):

spacy/tests/test_language.py

@@ -6,6 +6,7 @@ from spacy.vocab import Vocab
 from spacy.training import Example
 from spacy.lang.en import English
 from spacy.util import registry
+import spacy
 
 from .util import add_vecs_to_vocab, assert_docs_equal
 
@@ -266,3 +267,13 @@ def test_language_custom_tokenizer():
     assert [t.text for t in doc] == ["_hello", "_world"]
     doc = list(nlp.pipe(["hello world"]))[0]
     assert [t.text for t in doc] == ["_hello", "_world"]
+
+
+def test_spacy_blank():
+    nlp = spacy.blank("en")
+    assert nlp.config["training"]["dropout"] == 0.1
+    config = {"training": {"dropout": 0.2}}
+    meta = {"name": "my_custom_model"}
+    nlp = spacy.blank("en", config=config, meta=meta)
+    assert nlp.config["training"]["dropout"] == 0.2
+    assert nlp.meta["name"] == "my_custom_model"