Update chatbot_rag.py

chatbot_rag.py

@@ -62,7 +62,7 @@ def build_qa():
         tokenizer=tokenizer,
         max_new_tokens=256,       # control length of response
         temperature=0.2,          # more deterministic
-        do_sample=False,          # no randomness (deterministic answers)
+        do_sample=True,          # no randomness (deterministic answers)
         top_p=0.9,                # nucleus sampling
         repetition_penalty=1.2,   # 🚀 reduce loops/repeats
         eos_token_id=tokenizer.eos_token_id,