Python Library for DLAB utils
You can install the dlabutils
package using pip:
pip install git+https://github.com/epfl-dlab/dlab-utils.git
The model_path
function returns the path to the model weights given the huggingface identifier, if the model is available in the DLAB LLM weights directory, and defaults to the hf model name if it is not. The available_models
function returns a list of all available models. Here's a quick tutorial:
from dlabutils import model_path, available_models
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained(model_path("meta-llama/Meta-Llama-3-8B-Instruct"))
tokenizer = AutoTokenizer.from_pretrained(model_path("meta-llama/Meta-Llama-3-8B-Instruct"))
print(available_models())