fix: added

server.js

@@ -97,6 +97,7 @@ async function runTextInference(prompt, maxTokens) {
   const output = await model.generate({
     ...inputs,
     max_new_tokens: maxTokens,
+    do_sample: false,
   });
 
   const promptLength = inputs.input_ids.dims.at(-1);
@@ -284,6 +285,7 @@ app.post("/prompt", express.json(), async (req, res) => {
       request_id: req.requestId,
       duration_ms: Date.now() - start,
       response_chars: response?.length ?? 0,
+      model_output: response,
     });
     res.json({ response });
   } catch (err) {