Don't allow empty prompts. Context past always equal or greater than zero.

llm.cpp

@@ -67,6 +67,7 @@ bool LLMObject::isModelLoaded() const
 void LLMObject::resetResponse()
 {
     s_ctx.n_past -= m_responseTokens;
+    s_ctx.n_past = std::max(0, s_ctx.n_past);
     s_ctx.logits.erase(s_ctx.logits.end() -= m_responseLogits, s_ctx.logits.end());
     m_responseTokens = 0;
     m_responseLogits = 0;

main.qml

@@ -713,6 +713,9 @@ Window {
                 radius: 10
             }
             onAccepted: {
+                if (textInput.text === "")
+                    return
+
                 LLM.stopGenerating()
 
                 if (chatModel.count) {