In the dynamic landscape of generative NLP, traditional text processing pipelines limit research flexibility and reproducibility, as they are tailored to specific dataset, task, and model combinations. The escalating complexity, involving system prompts, model-specific formats, instructions, and more, calls for a shift to a structured, modular, and customizable solution.
Addressing this need, we present Unitxt, an innovative library for customizable textual data preparation and evaluation tailored to generative language models. Unitxt natively integrates with common libraries like HuggingFace and LM-eval-harness and deconstructs processing flows into modular components, enabling easy customization and sharing between practitioners. These components encompass model-specific formats, task prompts, and many other comprehensive dataset processing definitions. The Unitxt-Catalog centralizes these components, fostering collaboration and exploration in modern textual data workflows. Beyond being a tool, Unitxt is a community-driven platform, empowering users to build, share, and advance their pipelines collaboratively.
Turn.on.speakers.mov
To launch unitxt graphical user interface first install unitxt with ui requirements:
pip install unitxt[ui]
Then launch the ui by running:
unitxt-explore
This is a simple example of running end-to-end evaluation in self contained python code over user data.
See more examples in examples subdirectory.
from unitxt import get_logger
from unitxt.api import evaluate, load_dataset
from unitxt.blocks import Task, TaskCard
from unitxt.inference import HFPipelineBasedInferenceEngine
from unitxt.loaders import LoadFromDictionary
from unitxt.templates import InputOutputTemplate, TemplatesDict
from unitxt.text_utils import print_dict
logger = get_logger()
# Set up question answer pairs in a dictionary
data = {
"test": [
{"question": "What is the capital of Texas?", "answer": "Austin"},
{"question": "What is the color of the sky?", "answer": "Blue"},
]
}
card = TaskCard(
# Load the data from the dictionary. Data can be also loaded from HF, CSV files, COS and other sources using different loaders.
loader=LoadFromDictionary(data=data),
# Define the QA task input and output and metrics.
task=Task(
input_fields={"question": str},
reference_fields={"answer": str},
prediction_type=str,
metrics=["metrics.accuracy"],
),
)
# Create a simple template that formats the input.
# Add lowercase normalization as a post processor on the model prediction.
template = InputOutputTemplate(
instruction="Answer the following question.",
input_format="{question}",
output_format="{answer}",
postprocessors=["processors.lower_case"],
)
# Verbalize the dataset using the template
dataset = load_dataset(card=card, template=template)
test_dataset = dataset["test"]
# Infer using flan t5 base using HF API
# can be replaced with any prediction code,
# including the built in WMLInferenceEngine and OpenAiInferenceEngine.
model_name = "google/flan-t5-base"
inference_model = HFPipelineBasedInferenceEngine(
model_name=model_name, max_new_tokens=32
)
predictions = inference_model.infer(test_dataset)
evaluated_dataset = evaluate(predictions=predictions, data=test_dataset)
# Print results
for instance in evaluated_dataset:
print_dict(
instance,
keys_to_print=[
"source", # input to the model
"prediction", # model prediction
"processed_prediction", # model prediction after post processing
"references", # reference answer
"score", # scores (per instance and global)
],
)
Please install Unitxt from source by:
git clone git@github.com:IBM/unitxt.git
cd unitxt
pip install -e ".[dev]"
pre-commit install
If you use Unitxt in your research, please cite our paper:
@inproceedings{bandel-etal-2024-unitxt,
title = "Unitxt: Flexible, Shareable and Reusable Data Preparation and Evaluation for Generative {AI}",
author = "Bandel, Elron and
Perlitz, Yotam and
Venezian, Elad and
Friedman, Roni and
Arviv, Ofir and
Orbach, Matan and
Don-Yehiya, Shachar and
Sheinwald, Dafna and
Gera, Ariel and
Choshen, Leshem and
Shmueli-Scheuer, Michal and
Katz, Yoav",
editor = "Chang, Kai-Wei and
Lee, Annie and
Rajani, Nazneen",
booktitle = "Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 3: System Demonstrations)",
month = jun,
year = "2024",
address = "Mexico City, Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.naacl-demo.21",
pages = "207--215",
abstract = "In the dynamic landscape of generative NLP, traditional text processing pipelines limit research flexibility and reproducibility, as they are tailored to specific dataset, task, and model combinations. The escalating complexity, involving system prompts, model-specific formats, instructions, and more, calls for a shift to a structured, modular, and customizable solution.Addressing this need, we present Unitxt, an innovative library for customizable textual data preparation and evaluation tailored to generative language models. Unitxt natively integrates with common libraries like HuggingFace and LM-eval-harness and deconstructs processing flows into modular components, enabling easy customization and sharing between practitioners. These components encompass model-specific formats, task prompts, and many other comprehensive dataset processing definitions. The Unitxt Catalog centralizes these components, fostering collaboration and exploration in modern textual data workflows. Beyond being a tool, Unitxt is a community-driven platform, empowering users to build, share, and advance their pipelines collaboratively. Join the Unitxt community at https://github.com/IBM/unitxt",
}
Unitxt emoji designed by OpenMoji - the open-source emoji and icon project. License: CC BY-SA 4.0
