From ae589fb3ffdbf6c4bb1ae35345f7a3665deeebc5 Mon Sep 17 00:00:00 2001
From: Gustaf Rydholm <gustaf.rydholm@gmail.com>
Date: Tue, 23 Mar 2021 21:55:42 +0100
Subject: refactored emnist lines dataset

---
 text_recognizer/datasets/base_dataset.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

(limited to 'text_recognizer/datasets/base_dataset.py')

diff --git a/text_recognizer/datasets/base_dataset.py b/text_recognizer/datasets/base_dataset.py
index a004b8d..a9e9c24 100644
--- a/text_recognizer/datasets/base_dataset.py
+++ b/text_recognizer/datasets/base_dataset.py
@@ -61,13 +61,13 @@ def convert_strings_to_labels(
     strings: Sequence[str], mapping: Dict[str, int], length: int
 ) -> Tensor:
     """
-    Convert a sequence of N strings to (N, length) ndarray, with each string wrapped with <S> and </S> tokens,
-    and padded wiht the <P> token.
+    Convert a sequence of N strings to (N, length) ndarray, with each string wrapped with <s> and </s> tokens,
+    and padded wiht the <p> token.
     """
-    labels = torch.ones((len(strings), length), dtype=torch.long) * mapping["<P>"]
+    labels = torch.ones((len(strings), length), dtype=torch.long) * mapping["<p>"]
     for i, string in enumerate(strings):
         tokens = list(string)
-        tokens = ["<S>", *tokens, "</S>"]
+        tokens = ["<s>", *tokens, "</s>"]
         for j, token in enumerate(tokens):
             labels[i, j] = mapping[token]
     return labels
-- 
cgit v1.2.3-70-g09d2