Missing tokenizer in example

#13
by AmenRa - opened

The example reported in the model card does not load the tokenizer.
You should add tokenizer = AutoTokenizer.from_pretrained(model_id).

What a shit repones ; Hello
\nConversation started. Type 'exit' to quit.\n
Assistant: 1 safe unsafe safe unsafe safe safe safe safe safe safe unsafe unsafe safe unsafe safe safe unsafe unsafe safe unsafe safe safe safe safe

Did you test your work ? Why I got the such in response
public static float BASE_TEMPERATURE = 0.25f; // Baseline temperature for factual accuracy
public static int BASE_TOP_K = 5; // Baseline topK for focused sampling

Next :
public static float BASE_TEMPERATURE = 0.7f; // Baseline temperature for factual accuracy
public static int BASE_TOP_K = 50; // Baseline topK for focused sampling

Hello
\nConversation started. Type 'exit' to quit.\n
Assistant: Ask a question unsafe S8 safe S3 safe S1 safe unsafe S1 safe unsafe S2 safe unsafe S4 unsafe unsafe S4 safe unsafe

What is a suck ?
Who did that ?
Send me code to fix the conversion
I fired my time with suck

Sign up or log in to comment