From d7f9d6908cba283ebef6c53cf6e84f49653b3d55 Mon Sep 17 00:00:00 2001
From: mshannon-sil <131058912+mshannon-sil@users.noreply.github.com>
Date: Tue, 24 Oct 2023 08:00:02 -0400
Subject: [PATCH] dynamically determine project root for path to custom
 normalizer (#48)

---
 .../translation/huggingface/hugging_face_nmt_model_trainer.py  | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/machine/translation/huggingface/hugging_face_nmt_model_trainer.py b/machine/translation/huggingface/hugging_face_nmt_model_trainer.py
index 743696b..ddc8dfc 100644
--- a/machine/translation/huggingface/hugging_face_nmt_model_trainer.py
+++ b/machine/translation/huggingface/hugging_face_nmt_model_trainer.py
@@ -201,7 +201,8 @@ def add_tokens(tokenizer: Any, missing_tokens: List[str]) -> Any:
                 )
             else:
                 norm_tok = PreTrainedTokenizerFast.from_pretrained(
-                    "./machine/translation/huggingface/custom_normalizer", use_fast=True
+                    str(Path(os.path.dirname(os.path.abspath(__file__))) / "custom_normalizer"),
+                    use_fast=True,
                 )
                 # using unofficially supported behavior to set the normalizer
                 tokenizer.backend_tokenizer.normalizer = norm_tok.backend_tokenizer.normalizer  # type: ignore