diff --git a/torchtext/csrc/vocab.cpp b/torchtext/csrc/vocab.cpp
index 1831d46f39..970573713b 100644
--- a/torchtext/csrc/vocab.cpp
+++ b/torchtext/csrc/vocab.cpp
@@ -1,4 +1,5 @@
 #include <ATen/Parallel.h> // @manual
+#include <algorithm>
 #include <common.h>
 #include <iostream>
 #include <stdexcept>
@@ -8,7 +9,8 @@
 namespace torchtext {
 
 Vocab::Vocab(const StringList &tokens, const std::string &unk_token)
-    : stoi_(MAX_VOCAB_SIZE, -1), unk_token_(std::move(unk_token)) {
+    : unk_token_(std::move(unk_token)) {
+  std::fill(stoi_.begin(), stoi_.end(), -1);
   for (std::size_t i = 0; i < tokens.size(); i++) {
     // tokens should not have any duplicates
     auto token_position =
diff --git a/torchtext/csrc/vocab.h b/torchtext/csrc/vocab.h
index d915c7de27..b9b17a3263 100644
--- a/torchtext/csrc/vocab.h
+++ b/torchtext/csrc/vocab.h
@@ -12,7 +12,7 @@ typedef std::tuple<std::string, std::vector<int64_t>, std::vector<std::string>,
 struct Vocab : torch::CustomClassHolder {
   static const int32_t MAX_VOCAB_SIZE = 30000000;
   int64_t unk_index_;
-  std::vector<int32_t> stoi_;
+  std::array<uint32_t, MAX_VOCAB_SIZE> stoi_;
   const std::string version_str_ = "0.0.1";
   StringList itos_;
   std::string unk_token_;
@@ -44,7 +44,7 @@ struct Vocab : torch::CustomClassHolder {
   uint32_t _find(const c10::string_view &w) const {
     uint32_t stoi_size = stoi_.size();
     uint32_t id = _hash(w) % stoi_size;
-    while (stoi_[id] != -1 && itos_[stoi_[id]]!= w) {
+    while (stoi_[id] != -1 && itos_[stoi_[id]] != w) {
       id = (id + 1) % stoi_size;
     }
     return id;