Disable caching during text generation

Author: rhymeswithlion

module-16-applying-peft-on-foundation-models/exercises/solution/teach-an-llm-to-spell-with-sft-solution.ipynb

@@ -335,7 +335,7 @@
     "\n",
     "    # Generate text from the model\n",
     "    gen = model.generate(\n",
-    "        **inputs, max_new_tokens=max_new_tokens\n",
+    "        **inputs, max_new_tokens=max_new_tokens, use_cache=False\n",
     "    )  # No parameters = greedy search\n",
     "\n",
     "    # Decode the generated tokens to a string\n",