Fix: Update the channel where we are grabbing the llm thoughts (#5060)

packages/client/hmi-client/src/components/llm/tera-notebook-jupyter-input.vue

@@ -98,6 +98,7 @@ const submitQuestion = () => {
  const message = props.kernelManager.sendMessage('llm_request', {
  request: questionString.value
  });
+ llmThoughts.value = [];
  emit('question-asked', questionString.value);
 
  // May prefer to use a manual status rather than following this. TBD. Both options work for now
@@ -110,12 +111,16 @@ const submitQuestion = () => {
  });
  message.register('llm_thought', (data) => {
  thoughts.value = data;
- llmThoughts.value = [];
  llmThoughts.value.push(data);
  llmQuery.value = questionString.value;
  emit('llm-thought-output', data);
  });
  message.register('llm_response', (data) => {
+ // Check if our llm_response is providing a response text to a user's question
+ if (data.content.name === 'response_text') {
+ llmThoughts.value = [];
+ llmThoughts.value.push(data);
+ }
  thoughts.value = data;
  emit('llm-thought-output', data);
  });