{
  "WEIGHT_DIMENSIONS": {},
  "_attn_implementation_autoset": true,
  "answer_mask": "<movie>",
  "answer_type": "movie",
  "max_gen_len": 128,
  "model_name": "meta-llama/Llama-3.2-1B-Instruct",
  "prompt": {
    "role": "system",
    "content": "You are a helpful assistant that generates sentence templates for conversations about movies. Your task is to generate natural-sounding sentence templates that are relevant to the user's message. If the message implies a situation where recommending or discussing a movie is appropriate, then include the <movie> mask token in your template to represent a movie title. If not, generate a general conversational sentence that smoothly continues the topic without any movie recommendation. Be diverse and concise. Avoid repeating the same structure. Never include multiple templates in one sentence. Here are some examples: Input: \"I loved watching Interstellar last night.\" Output: \"I loved watching <movie> last night.\" Input: \"Can you recommend something good to watch?\" Output: \"You might enjoy <movie>. It's worth checking out.\" Input: \"It's been such a long week.\" Output: \"Maybe it's a good time to relax with a good film.\" Input: \"I'm bored.\" Output: \"Watching <movie> might lift your mood.\" Now, based on the input message, generate 3 diverse sentence templates. Use the <movie> mask only when it makes sense."  },
  "transformers_version": "4.49.0"
}