llama : fix bpe tokenize from byte (ggerganov#2889)

yxq321 · Sep 3, 2023 · 3730134 · 3730134
1 parent d9151e6
commit 3730134
Showing 1 changed file with 8 additions and 2 deletions.
diff --git a/llama.cpp b/llama.cpp
@@ -3366,9 +3366,15 @@ struct llm_tokenizer_bpe {
                         std::string byte_str(1, *j);
                         auto token_multibyte = vocab.token_to_id.find(byte_str);
                         if (token_multibyte == vocab.token_to_id.end()) {
-                            fprintf(stderr,"ERROR: byte not found in vocab: '%s'\n", byte_str.c_str());
+                            try {
+                                llama_token token_byte = llama_byte_to_token(vocab, *j);
+                                output.push_back(token_byte);
+                            } catch (const std::out_of_range & err) {
+                                fprintf(stderr,"ERROR: byte not found in vocab: '%s'\n", byte_str.c_str());
+                            }
+                        } else {
+                            output.push_back((*token_multibyte).second);
                         }
-                        output.push_back((*token_multibyte).second);
                     }
                 } else {
                     output.push_back((*token).second);