From 6cd62e3d488afc338dcbb0f5f0cda209480fd730 Mon Sep 17 00:00:00 2001
From: theblackcat102 <theblackcat102@github.com>
Date: Fri, 20 Jan 2023 06:16:26 +0000
Subject: [PATCH] [fix] Fix missing russian and update readme

---
 model/supervised_finetuning/README.md           | 17 +++++++++++++++++
 .../custom_datasets/translation.py              |  5 +++--
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/model/supervised_finetuning/README.md b/model/supervised_finetuning/README.md
index 822121d8..9f200847 100644
--- a/model/supervised_finetuning/README.md
+++ b/model/supervised_finetuning/README.md
@@ -60,6 +60,23 @@ python trainer.py --configs defaults your-model-name --deepspeed
 
 ## Dataset choices
 
+To specify which translation pair for [WMT](https://huggingface.co/datasets/wmt19) and [TED Talk](https://huggingface.co/datasets/ted_talks_iwslt) translation simply add the supported language pair at the postfix
+
+```
+  datasets:
+    - wmt2019_zh-en
+    - wmt2019_ru-en
+    - wmt2019_de-en
+    - ted_trans_nl-en
+    - ted_trans_de-ja
+```
+
+Currently only these languages are supported via prompt translation:
+
+```
+ar,de,fr,en,it,nl,tr,ru,ms,ko,ja,zh
+```
+
 ## Results
 
 Experimental results in wandb
diff --git a/model/supervised_finetuning/custom_datasets/translation.py b/model/supervised_finetuning/custom_datasets/translation.py
index a6d46e9e..79fff0d1 100644
--- a/model/supervised_finetuning/custom_datasets/translation.py
+++ b/model/supervised_finetuning/custom_datasets/translation.py
@@ -63,10 +63,11 @@ TRANSLATION_PROMPT = {
         "{}, jemahkan ayat ini kepada bahasa melayu"
     ],
     "en": ["{}. translate to english", "{} write in english", "english translation: '{}'"],
-    "tr": ["{}. translate to turkish", "{} write in turkish", "turkish translation: '{}'"],
+    "ru": ["помогите мне перевести это на русский : {}", "{} перевести на русский язык", "russian translation: '{}'"],
+    "tr": ["{}. türkçeye çevi̇ri̇n", "{} write in turkish", "turkish translation: '{}'", "türkçeye çevi̇rmek: {}"],
     "it": ["{}. translate to italian", "{} write in italian", "italian translation: '{}'"],
     "nl": ["{}. translate to dutch", "{} write in dutch", "dutch translation: '{}'"],
-    "vi": ["{}. translate to vietnamese", "{} write in vietnamese", "vietnamese translation: '{}'"],
+    "vi": ["{}. Dịch sang tiếng việt nam", "{} write in vietnamese", "vietnamese translation: '{}'"],
     "ar": ["{}. translate to arabic", "{} write in arabic", "arabic translation: '{}'"],
 }
 class TranslationPair(Dataset):