From 1d7bbbffb6a024a33859b48a7a656d0455dc0be1 Mon Sep 17 00:00:00 2001
From: zhifu gao <zhifu.gzf@alibaba-inc.com>
Date: 星期一, 16 十月 2023 11:47:59 +0800
Subject: [PATCH] Update README.md
---
funasr/runtime/onnxruntime/src/vocab.cpp | 24 ++++++++++++++++++++++++
1 files changed, 24 insertions(+), 0 deletions(-)
diff --git a/funasr/runtime/onnxruntime/src/vocab.cpp b/funasr/runtime/onnxruntime/src/vocab.cpp
index 95174c7..2babc40 100644
--- a/funasr/runtime/onnxruntime/src/vocab.cpp
+++ b/funasr/runtime/onnxruntime/src/vocab.cpp
@@ -75,6 +75,21 @@
return false;
}
+string Vocab::WordFormat(std::string word)
+{
+ if(word == "i"){
+ return "I";
+ }else if(word == "i'm"){
+ return "I'm";
+ }else if(word == "i've"){
+ return "I've";
+ }else if(word == "i'll"){
+ return "I'll";
+ }else{
+ return word;
+ }
+}
+
string Vocab::Vector2StringV2(vector<int> in, std::string language)
{
int i;
@@ -94,6 +109,7 @@
size_t found = word.find(unicodeChar);
if(found != std::string::npos){
if (combine != ""){
+ combine = WordFormat(combine);
if (words.size() != 0){
combine = " " + combine;
}
@@ -163,6 +179,14 @@
}
}
+ if (language == "en-bpe" and combine != ""){
+ combine = WordFormat(combine);
+ if (words.size() != 0){
+ combine = " " + combine;
+ }
+ words.push_back(combine);
+ }
+
stringstream ss;
for (auto it = words.begin(); it != words.end(); it++) {
ss << *it;
--
Gitblit v1.9.1