From: Henri Vasserman <redacted>
Date: Mon, 3 Jul 2023 21:05:23 +0000 (+0300)
Subject: fix server crashes (#2076)
X-Git-Tag: gguf-v0.4.0~528
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=1cf14ccef12e19c5a5b0b17ab456242d1f8c7fdd;p=pkg%2Fggml%2Fsources%2Fllama.cpp

fix server crashes (#2076)
---

diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index e4ddbe98..3bf98595 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -906,7 +906,7 @@ int main(int argc, char ** argv) {
 
             while (llama.has_next_token) {
                 const completion_token_output token_with_probs = llama.doCompletion();
-                const std::string token_text = llama_token_to_str(llama.ctx, token_with_probs.tok);
+                const std::string token_text = token_with_probs.tok == -1 ? "" : llama_token_to_str(llama.ctx, token_with_probs.tok);
 
                 stop_pos = llama.findStoppingStrings(llama.generated_text,
                     token_text.size(), STOP_FULL);
@@ -933,7 +933,7 @@ int main(int argc, char ** argv) {
 
                 while (llama.has_next_token) {
                     const completion_token_output token_with_probs = llama.doCompletion();
-                    const std::string token_text = llama_token_to_str(llama.ctx, token_with_probs.tok);
+                    const std::string token_text = token_with_probs.tok == -1 ? "" : llama_token_to_str(llama.ctx, token_with_probs.tok);
                     if (llama.multibyte_pending > 0) {
                         continue;
                     }