Update src/generation/llm.py

src/generation/llm.py

@@ -93,12 +93,9 @@ class LLM:
             outputs = self.llm.generate(
                         **inputs,
                         max_new_tokens=max_length,
-                        temperature=0.3,
-                        do_sample=False,
-                        top_p=1.0,
-                        top_k=1,
+                        temperature=0.7,
+                        do_sample=True,
                         pad_token_id=self.tokenizer.eos_token_id,
-                        eos_token_id=self.tokenizer.eos_token_id,
                     )
             # Decode the generated tokens
             response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)