fix - thanks @joaogante !

README.md

@@ -96,7 +96,7 @@ inputs = processor(prompts, return_tensors="pt").to(device)
 # inputs = processor(prompts[0], return_tensors="pt").to(device)
 
 # Generation args
-bad_words_ids = tokenizer(["<image>", "<fake_token_around_image>"], add_special_tokens=False).input_ids
+bad_words_ids = processor.tokenizer(["<image>", "<fake_token_around_image>"], add_special_tokens=False).input_ids
 
 generated_ids = model.generate(**inputs, bad_words_ids=bad_words_ids, max_length=100)
 generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)
@@ -141,7 +141,7 @@ inputs = processor(prompts, add_end_of_utterance_token=False, return_tensors="pt
 
 # Generation args
 exit_condition = processor.tokenizer("<end_of_utterance>", add_special_tokens=False).input_ids
-bad_words_ids = tokenizer(["<image>", "<fake_token_around_image>"], add_special_tokens=False).input_ids
+bad_words_ids = processor.tokenizer(["<image>", "<fake_token_around_image>"], add_special_tokens=False).input_ids
 
 generated_ids = model.generate(**inputs, eos_token_id=exit_condition, bad_words_ids=bad_words_ids, max_length=100)
 generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)