tinystories_llama2-7b-chat_hf-endpoint.yaml

# Config for everything related to dataset generation inputs
input_config:
    # Dataset metadata
    domain: stories
    language: en

    # Dataset generator parameters
    quantity: 10
    
    # HuggingFace dataset params
    dataset: roneneldan/TinyStories
    dataset_text_column: text
    dataset_params:
        split: validation
    
    # Prompt template
    template: >-
        Write a fictional or creative story to entertain, evoke emotions, 
        or convey deeper meanings following the style of literature, film, 
        or theater. Base your story in the following entities: {entities}.\nStory: 
    
    # Extractor params
    extractor: entity_list
    max_input_tokens: 256

# Config for model instantiation
model_config:
    provider: hf_remote
    model_name: meta-llama/Llama-2-7b-chat-hf
    url: https://g6vjy620o2eplyzc.us-east-1.aws.endpoints.huggingface.cloud
    max_retries: 5
    threads: 8
    
# Decoding args
generation_config:
    # Ignore `max_new_tokens` or `min_new_tokens` to get  automatic length estimation
    # min_new_tokens: 5
    # max_new_tokens: 10
    do_sample: true
    top_p: 0.9
    temperature: 0.6
    repetition_penalty: 1.0
    return_full_text: false