From 1cf14ccef12e19c5a5b0b17ab456242d1f8c7fdd Mon Sep 17 00:00:00 2001 From: Henri Vasserman Date: Tue, 4 Jul 2023 00:05:23 +0300 Subject: [PATCH] fix server crashes (#2076) --- examples/server/server.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/examples/server/server.cpp b/examples/server/server.cpp index e4ddbe986..3bf985957 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; }