Fix up configuration keys

spacy/language.py

@@ -1837,7 +1837,7 @@ class Language:
         # using the nlp.config with all defaults.
         config = util.copy_config(config)
         orig_pipeline = config.pop("components", {})
-        orig_distill = config.pop("distill", None)
+        orig_distill = config.pop("distillation", None)
         orig_pretraining = config.pop("pretraining", None)
         config["components"] = {}
         if auto_fill:
@@ -1847,8 +1847,8 @@ class Language:
         filled["components"] = orig_pipeline
         config["components"] = orig_pipeline
         if orig_distill is not None:
-            filled["distill"] = orig_distill
-            config["distill"] = orig_distill
+            filled["distillation"] = orig_distill
+            config["distillation"] = orig_distill
         if orig_pretraining is not None:
             filled["pretraining"] = orig_pretraining
             config["pretraining"] = orig_pretraining

spacy/schemas.py

@@ -462,7 +462,7 @@ CONFIG_SCHEMAS = {
     "training": ConfigSchemaTraining,
     "pretraining": ConfigSchemaPretrain,
     "initialize": ConfigSchemaInit,
-    "distill": ConfigSchemaDistill,
+    "distillation": ConfigSchemaDistill,
 }
 
 

spacy/training/initialize.py

@@ -124,13 +124,11 @@ def init_nlp_distill(
     config = nlp.config.interpolate()
     # Resolve all training-relevant sections using the filled nlp config
     T = registry.resolve(config["training"], schema=ConfigSchemaTraining)
-    D = registry.resolve(config["distill"], schema=ConfigSchemaDistill)
+    D = registry.resolve(config["distillation"], schema=ConfigSchemaDistill)
     dot_names = [D["corpus"], T["dev_corpus"]]
     if not isinstance(D["corpus"], str):
         raise ConfigValidationError(
-            desc=Errors.E897.format(
-                field="distill.corpus", type=type(D["corpus"])
-            )
+            desc=Errors.E897.format(field="distillation.corpus", type=type(D["corpus"]))
         )
     if not isinstance(T["dev_corpus"], str):
         raise ConfigValidationError(
@@ -158,7 +156,9 @@ def init_nlp_distill(
     labels = {}
     for name, pipe in nlp.pipeline:
         # Copy teacher labels.
-        teacher_pipe_name = student_to_teacher[name] if name in student_to_teacher else name
+        teacher_pipe_name = (
+            student_to_teacher[name] if name in student_to_teacher else name
+        )
         teacher_pipe = teacher_pipes.get(teacher_pipe_name, None)
         if (
             teacher_pipe is not None

spacy/training/loop.py

@@ -57,7 +57,7 @@ def distill(
     if use_gpu >= 0 and allocator:
         set_gpu_allocator(allocator)
     T = registry.resolve(config["training"], schema=ConfigSchemaTraining)
-    D = registry.resolve(config["distill"], schema=ConfigSchemaDistill)
+    D = registry.resolve(config["distillation"], schema=ConfigSchemaDistill)
     dot_names = [D["corpus"], T["dev_corpus"]]
     distill_corpus, dev_corpus = resolve_dot_names(config, dot_names)
     optimizer = D["optimizer"]
@@ -333,7 +333,7 @@ def _distill_loop(
         if before_update:
             before_update_args = {"step": step, "epoch": epoch}
             before_update(student, before_update_args)
-        dropout = dropouts(optimizer.step) 
+        dropout = dropouts(optimizer.step)
         for subbatch in subdivide_batch(batch, accumulate_gradient):
             student.distill(
                 teacher,