From 3b5cfec1fcf34e45d86fd2b133120be13141488a Mon Sep 17 00:00:00 2001
From: Matthw Honnibal <honnibal+gh@gmail.com>
Date: Thu, 21 May 2020 19:32:04 +0200
Subject: [PATCH] Tweak memory management in train_from_config

---
 spacy/cli/train_from_config.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/spacy/cli/train_from_config.py b/spacy/cli/train_from_config.py
index c75c861cc..eeb21c10c 100644
--- a/spacy/cli/train_from_config.py
+++ b/spacy/cli/train_from_config.py
@@ -213,6 +213,12 @@ def train_from_config(
                 if is_best_checkpoint and output_path is not None:
                     nlp.to_disk(output_path)
                 progress = tqdm.tqdm(total=training["eval_frequency"], leave=False)
+            # Clean up the objects to faciliate garbage collection.
+            for eg in batch:
+                eg.doc = None
+                eg.goldparse = None
+                eg.doc_annotation = None
+                eg.token_annotation = None
     finally:
         if output_path is not None:
             final_model_path = output_path / "model-final"