LAI-Paca-7b / README.md
Artples's picture
Update README.md
f46f7a0 verified
metadata
tags:
  - text-generation-inference
  - text-generation
  - peft
library_name: transformers
widget:
  - messages:
      - role: user
        content: >-
          Translate the text 'Bonjour, comment allez-vous?' from French to
          English.
datasets:
  - tatsu-lab/alpaca
metrics:
  - accuracy
pipeline_tag: text-generation
license: cc-by-nc-4.0
inference: false

LAI-Paca-7b

Instruction Fine-Tune of Mistral with the Alpaca dataset for instructions. It should primarily be used for API calls for tools, such as making a call to obtain a title for a song generated by a music generation AI based on a provided prompt or other applications.

Notice

Please remember that the uploaded model is a adapter model.
It should be used in the Alpaca format.

Usage


from transformers import AutoModelForCausalLM, AutoTokenizer

model_path = "Artples/LAI-Paca-7b"

tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    device_map="auto",
    torch_dtype='auto'
).eval()

# Prompt content: "hi"
messages = [
    {"role": "user", "content": "hi"}
]

input_ids = tokenizer.apply_chat_template(conversation=messages, tokenize=True, add_generation_prompt=True, return_tensors='pt')
output_ids = model.generate(input_ids.to('cuda'))
response = tokenizer.decode(output_ids[0][input_ids.shape[1]:], skip_special_tokens=True)

# Model response: "Hello! How can I assist you today?"
print(response)