NeMo-Curator
Scalable data pre processing and curation toolkit for LLMs
Stars: 438
NeMo Curator is a GPU-accelerated open-source framework designed for efficient large language model data curation. It provides scalable dataset preparation for tasks like foundation model pretraining, domain-adaptive pretraining, supervised fine-tuning, and parameter-efficient fine-tuning. The library leverages GPUs with Dask and RAPIDS to accelerate data curation, offering customizable and modular interfaces for pipeline expansion and model convergence. Key features include data download, text extraction, quality filtering, deduplication, downstream-task decontamination, distributed data classification, and PII redaction. NeMo Curator is suitable for curating high-quality datasets for large language model training.
README:
🚀 The GPU-Accelerated Open Source Framework for Efficient Large Language Model Data Curation 🚀
NeMo Curator is a Python library specifically designed for fast and scalable dataset preparation and curation for large language model (LLM) use-cases such as foundation model pretraining, domain-adaptive pretraining (DAPT), supervised fine-tuning (SFT) and paramter-efficient fine-tuning (PEFT). It greatly accelerates data curation by leveraging GPUs with Dask and RAPIDS, resulting in significant time savings. The library provides a customizable and modular interface, simplifying pipeline expansion and accelerating model convergence through the preparation of high-quality tokens.
At the core of the NeMo Curator is the DocumentDataset
which serves as the the main dataset class. It acts as a straightforward wrapper around a Dask DataFrame
. The Python library offers easy-to-use methods for expanding the functionality of your curation pipeline while eliminating scalability concerns.
NeMo Curator provides a collection of scalable data-mining modules. Some of the key features include:
-
Data download and text extraction
- Default implementations for downloading and extracting Common Crawl, Wikipedia, and ArXiv data
- Easily customize the download and extraction and extend to other datasets
-
Language identification and separation with fastText and pycld2
-
Text reformatting and cleaning to fix unicode decoding errors via ftfy
-
- Multilingual heuristic-based filtering
- Classifier-based filtering via fastText
-
- exact and fuzzy (near-identical) deduplication are accelerated using cuDF and Dask
- For fuzzy deduplication, our implementation follows the method described in Microsoft Turing NLG 530B
- For semantic deduplication, our implementation follows the method described in SemDeDup by Meta AI (FAIR) facebookresearch/SemDeDup
-
Multilingual downstream-task decontamination following the approach of OpenAI GPT3 and Microsoft Turing NLG 530B
-
Distributed data classification
- Multi-node, multi-GPU classifier inference
- Provides sophisticated domain and quality classification
- Flexible interface for extending to your own classifier network
-
Personal identifiable information (PII) redaction for removing addresses, credit card numbers, social security numbers, and more
These modules offer flexibility and permit reordering, with only a few exceptions. In addition, the NeMo Framework Launcher provides pre-built pipelines that can serve as a foundation for your customization use cases.
- Documentation
- Examples
- Tutorials
- Blog posts
- Curating Trillion-Token Datasets: Introducing NVIDIA NeMo Data Curator
- Scale and Curate High-Quality Datasets for LLM Training with NVIDIA NeMo Curator
- Curating Custom Datasets for LLM Training with NVIDIA NeMo Curator
- Curating Custom Datasets for LLM Parameter-Efficient Fine-Tuning with NVIDIA NeMo Curator
This section explains how to install NeMo Curator and use the Python library, Python modules, and CLI scripts. It also includes a list of tutorials to help you get started right away. Finally, this section explains how to use the NeMo Framework Launcher as an alternative method for interfacing with NeMo Curator.
Before installing NeMo Curator, ensure that the following requirements are met:
- Python 3.10
- Ubuntu 22.04/20.04
- NVIDIA GPU (optional)
- Volta™ or higher (compute capability 7.0+)
- CUDA 12 (or above)
You can install NeMo-Curator from PyPi, from source or get it through the NeMo Framework container.
To install the CPU-only modules:
pip install nemo-curator
To install the CPU and CUDA-accelerated modules:
pip install --extra-index-url https://pypi.nvidia.com nemo-curator[cuda12x]
-
Clone the NeMo Curator repository in GitHub.
git clone https://github.com/NVIDIA/NeMo-Curator.git cd NeMo-Curator
-
Install the modules that you need.
To install the CPU-only modules:
pip install .
To install the CPU and CUDA-accelerated modules:
pip install --extra-index-url https://pypi.nvidia.com ".[cuda12x]"
The latest release of NeMo Curator comes preinstalled in the NeMo Framework Container. If you want the latest commit inside the container, you can reinstall NeMo Curator using:
pip uninstall nemo-curator
rm -r /opt/NeMo-Curator
git clone https://github.com/NVIDIA/NeMo-Curator.git /opt/NeMo-Curator
pip install --extra-index-url https://pypi.nvidia.com /opt/NeMo-Curator[cuda12x]
And follow the instructions for installing from source from above.
The following snippet demonstrates how to create a small data curation pipeline that downloads and curates a small subset of the Common Crawl dataset.
# Download your dataset
dataset = download_common_crawl("/datasets/common_crawl/", "2021-04", "2021-10", url_limit=10)
# Build your pipeline
curation_pipeline = Sequential([
# Fix unicode
Modify(UnicodeReformatter()),
# Discard short records
ScoreFilter(WordCountFilter(min_words=80)),
# Discard low-quality records
ScoreFilter(FastTextQualityFilter(model_path="model.bin")),
# Discard records from the evaluation metrics to prevent test set leakage.
TaskDecontamination([Winogrande(), Squad(), TriviaQA()])
])
# Execute the pipeline on your dataset
curated_dataset = curation_pipeline(dataset)
To get started with NeMo Curator, you can follow the tutorials available here. These tutorials include:
-
tinystories
which focuses on data curation for training LLMs from scratch. -
peft-curation
which focuses on data curation for LLM parameter-efficient fine-tuning (PEFT) use-cases. -
distributed_data_classification
which focuses on using the quality and domain classifiers to help with data annotation. -
single_node_tutorial
which demonstrates an end-to-end data curation pipeline for curating Wikipedia data in Thai.
The NeMo Curator section of the NeMo Framework User Guide provides in-depth information about how the Python modules work. The examples directory in the GitHub repository provides scripts that showcase these modules.
NeMo Curator also offers CLI scripts for you to use. The scripts in nemo_curator/scripts
map closely to the supplied Python modules. Refer to the NeMo Framework User Guide for more information about the Python modules and scripts.
As an alternative method for interfacing with NeMo Curator, you can use the NeMo Framework Launcher. The launcher enables you to easily configure the parameters and cluster. It can also automatically generate the SLURM batch scripts that wrap around the CLI scripts required to run your pipeline.
In addition, other methods are available to run NeMo Curator on SLURM. For example, refer to the example scripts in examples/slurm
for information on how to run NeMo Curator on SLURM without the NeMo Framework Launcher.
The modules within NeMo Curator were primarily designed to curate high-quality documents from Common Crawl snapshots in a scalable manner. To evaluate the quality of the curated Common Crawl documents, we conducted a series of ablation experiments. In these experiments, we trained a 357M-parameter GPT-style model using datasets generated at various stages of our data curation pipeline, which was implemented in NeMo Curator.
The following figure shows that the use of different data curation modules implemented in NeMo Curator led to improved model zero-shot downstream task performance.
In terms of scalability and compute performance, using the combination of RAPIDS and Dask fuzzy deduplication enabled us to deduplicate the 1.1 Trillion token Red Pajama dataset in 1.8 hours with 64 NVIDIA A100 Tensor Core GPUs.
Additionally, using the CPU-based modules, the following table shows the time required and resulting data size reduction for each processing step Common Crawl snapshot from November/December of 2020 using 30 CPU nodes (with hardware similar to the c5.24xlarge
Amazon AWS C5 instance).
Dataset | Download and text extraction | Text cleaning | Quality filtering | |||
---|---|---|---|---|---|---|
Time | Output Size | Time | Output Size | Time | Output Size | |
Common Crawl 2020-50 | 36 hrs | 2.8 TB | 1 hr | 2.8 TB | 0.2 hr | 0.52 TB |
We welcome community contributions! Please refer to CONTRIBUTING.md for the process.
For Tasks:
Click tags to check more tools for each tasksFor Jobs:
Alternative AI tools for NeMo-Curator
Similar Open Source Tools
NeMo-Curator
NeMo Curator is a GPU-accelerated open-source framework designed for efficient large language model data curation. It provides scalable dataset preparation for tasks like foundation model pretraining, domain-adaptive pretraining, supervised fine-tuning, and parameter-efficient fine-tuning. The library leverages GPUs with Dask and RAPIDS to accelerate data curation, offering customizable and modular interfaces for pipeline expansion and model convergence. Key features include data download, text extraction, quality filtering, deduplication, downstream-task decontamination, distributed data classification, and PII redaction. NeMo Curator is suitable for curating high-quality datasets for large language model training.
aimo-progress-prize
This repository contains the training and inference code needed to replicate the winning solution to the AI Mathematical Olympiad - Progress Prize 1. It consists of fine-tuning DeepSeekMath-Base 7B, high-quality training datasets, a self-consistency decoding algorithm, and carefully chosen validation sets. The training methodology involves Chain of Thought (CoT) and Tool Integrated Reasoning (TIR) training stages. Two datasets, NuminaMath-CoT and NuminaMath-TIR, were used to fine-tune the models. The models were trained using open-source libraries like TRL, PyTorch, vLLM, and DeepSpeed. Post-training quantization to 8-bit precision was done to improve performance on Kaggle's T4 GPUs. The project structure includes scripts for training, quantization, and inference, along with necessary installation instructions and hardware/software specifications.
openvino
OpenVINO™ is an open-source toolkit for optimizing and deploying AI inference. It provides a common API to deliver inference solutions on various platforms, including CPU, GPU, NPU, and heterogeneous devices. OpenVINO™ supports pre-trained models from Open Model Zoo and popular frameworks like TensorFlow, PyTorch, and ONNX. Key components of OpenVINO™ include the OpenVINO™ Runtime, plugins for different hardware devices, frontends for reading models from native framework formats, and the OpenVINO Model Converter (OVC) for adjusting models for optimal execution on target devices.
nixtla
Nixtla is a production-ready generative pretrained transformer for time series forecasting and anomaly detection. It can accurately predict various domains such as retail, electricity, finance, and IoT with just a few lines of code. TimeGPT introduces a paradigm shift with its standout performance, efficiency, and simplicity, making it accessible even to users with minimal coding experience. The model is based on self-attention and is independently trained on a vast time series dataset to minimize forecasting error. It offers features like zero-shot inference, fine-tuning, API access, adding exogenous variables, multiple series forecasting, custom loss function, cross-validation, prediction intervals, and handling irregular timestamps.
Geoweaver
Geoweaver is an in-browser software that enables users to easily compose and execute full-stack data processing workflows using online spatial data facilities, high-performance computation platforms, and open-source deep learning libraries. It provides server management, code repository, workflow orchestration software, and history recording capabilities. Users can run it from both local and remote machines. Geoweaver aims to make data processing workflows manageable for non-coder scientists and preserve model run history. It offers features like progress storage, organization, SSH connection to external servers, and a web UI with Python support.
airavata
Apache Airavata is a software framework for executing and managing computational jobs on distributed computing resources. It supports local clusters, supercomputers, national grids, academic and commercial clouds. Airavata utilizes service-oriented computing, distributed messaging, and workflow composition. It includes a server package with an API, client SDKs, and a general-purpose UI implementation called Apache Airavata Django Portal.
vertex-ai-mlops
Vertex AI is a platform for end-to-end model development. It consist of core components that make the processes of MLOps possible for design patterns of all types.
habitat-lab
Habitat-Lab is a modular high-level library for end-to-end development in embodied AI. It is designed to train agents to perform a wide variety of embodied AI tasks in indoor environments, as well as develop agents that can interact with humans in performing these tasks.
generative-bi-using-rag
Generative BI using RAG on AWS is a comprehensive framework designed to enable Generative BI capabilities on customized data sources hosted on AWS. It offers features such as Text-to-SQL functionality for querying data sources using natural language, user-friendly interface for managing data sources, performance enhancement through historical question-answer ranking, and entity recognition. It also allows customization of business information, handling complex attribution analysis problems, and provides an intuitive question-answering UI with a conversational approach for complex queries.
Advanced-QA-and-RAG-Series
This repository contains advanced LLM-based chatbots for Retrieval Augmented Generation (RAG) and Q&A with different databases. It provides guides on using AzureOpenAI and OpenAI API for each project. The projects include Q&A and RAG with SQL and Tabular Data, and KnowledgeGraph Q&A and RAG with Tabular Data. Key notes emphasize the importance of good column names, read-only database access, and familiarity with query languages. The chatbots allow users to interact with SQL databases, CSV, XLSX files, and graph databases using natural language.
pytorch-forecasting
PyTorch Forecasting is a PyTorch-based package for time series forecasting with state-of-the-art network architectures. It offers a high-level API for training networks on pandas data frames and utilizes PyTorch Lightning for scalable training on GPUs and CPUs. The package aims to simplify time series forecasting with neural networks by providing a flexible API for professionals and default settings for beginners. It includes a timeseries dataset class, base model class, multiple neural network architectures, multi-horizon timeseries metrics, and hyperparameter tuning with optuna. PyTorch Forecasting is built on pytorch-lightning for easy training on various hardware configurations.
bisheng
Bisheng is a leading open-source **large model application development platform** that empowers and accelerates the development and deployment of large model applications, helping users enter the next generation of application development with the best possible experience.
AITemplate
AITemplate (AIT) is a Python framework that transforms deep neural networks into CUDA (NVIDIA GPU) / HIP (AMD GPU) C++ code for lightning-fast inference serving. It offers high performance close to roofline fp16 TensorCore (NVIDIA GPU) / MatrixCore (AMD GPU) performance on major models. AITemplate is unified, open, and flexible, supporting a comprehensive range of fusions for both GPU platforms. It provides excellent backward capability, horizontal fusion, vertical fusion, memory fusion, and works with or without PyTorch. FX2AIT is a tool that converts PyTorch models into AIT for fast inference serving, offering easy conversion and expanded support for models with unsupported operators.
contracts
AXONE Smart Contracts repository hosts Smart Contracts for the AXONE network, compatible with any Cosmos blockchains using the CosmWasm framework. It includes storage, sovereignty, and resource management oriented Smart Contracts. Each contract has different functionalities and maturity stages, with detailed tech documentation and emojis indicating maturity levels. The repository provides tools for building, testing, deploying, and interacting with Smart Contracts, along with guidelines for contributing and community engagement.
superlinked
Superlinked is a compute framework for information retrieval and feature engineering systems, focusing on converting complex data into vector embeddings for RAG, Search, RecSys, and Analytics stack integration. It enables custom model performance in machine learning with pre-trained model convenience. The tool allows users to build multimodal vectors, define weights at query time, and avoid postprocessing & rerank requirements. Users can explore the computational model through simple scripts and python notebooks, with a future release planned for production usage with built-in data infra and vector database integrations.
RD-Agent
RD-Agent is a tool designed to automate critical aspects of industrial R&D processes, focusing on data-driven scenarios to streamline model and data development. It aims to propose new ideas ('R') and implement them ('D') automatically, leading to solutions of significant industrial value. The tool supports scenarios like Automated Quantitative Trading, Data Mining Agent, Research Copilot, and more, with a framework to push the boundaries of research in data science. Users can create a Conda environment, install the RDAgent package from PyPI, configure GPT model, and run various applications for tasks like quantitative trading, model evolution, medical prediction, and more. The tool is intended to enhance R&D processes and boost productivity in industrial settings.
For similar tasks
superagent
Superagent is an open-source AI assistant framework and API that allows developers to add powerful AI assistants to their applications. These assistants use large language models (LLMs), retrieval augmented generation (RAG), and generative AI to help users with a variety of tasks, including question answering, chatbot development, content generation, data aggregation, and workflow automation. Superagent is backed by Y Combinator and is part of YC W24.
Awesome-Segment-Anything
Awesome-Segment-Anything is a powerful tool for segmenting and extracting information from various types of data. It provides a user-friendly interface to easily define segmentation rules and apply them to text, images, and other data formats. The tool supports both supervised and unsupervised segmentation methods, allowing users to customize the segmentation process based on their specific needs. With its versatile functionality and intuitive design, Awesome-Segment-Anything is ideal for data analysts, researchers, content creators, and anyone looking to efficiently extract valuable insights from complex datasets.
simpletransformers
Simple Transformers is a library based on the Transformers library by HuggingFace, allowing users to quickly train and evaluate Transformer models with only 3 lines of code. It supports various tasks such as Information Retrieval, Language Models, Encoder Model Training, Sequence Classification, Token Classification, Question Answering, Language Generation, T5 Model, Seq2Seq Tasks, Multi-Modal Classification, and Conversational AI.
smile
Smile (Statistical Machine Intelligence and Learning Engine) is a comprehensive machine learning, NLP, linear algebra, graph, interpolation, and visualization system in Java and Scala. It covers every aspect of machine learning, including classification, regression, clustering, association rule mining, feature selection, manifold learning, multidimensional scaling, genetic algorithms, missing value imputation, efficient nearest neighbor search, etc. Smile implements major machine learning algorithms and provides interactive shells for Java, Scala, and Kotlin. It supports model serialization, data visualization using SmilePlot and declarative approach, and offers a gallery showcasing various algorithms and visualizations.
pgai
pgai simplifies the process of building search and Retrieval Augmented Generation (RAG) AI applications with PostgreSQL. It brings embedding and generation AI models closer to the database, allowing users to create embeddings, retrieve LLM chat completions, reason over data for classification, summarization, and data enrichment directly from within PostgreSQL in a SQL query. The tool requires an OpenAI API key and a PostgreSQL client to enable AI functionality in the database. Users can install pgai from source, run it in a pre-built Docker container, or enable it in a Timescale Cloud service. The tool provides functions to handle API keys using psql or Python, and offers various AI functionalities like tokenizing, detokenizing, embedding, chat completion, and content moderation.
NeMo-Curator
NeMo Curator is a GPU-accelerated open-source framework designed for efficient large language model data curation. It provides scalable dataset preparation for tasks like foundation model pretraining, domain-adaptive pretraining, supervised fine-tuning, and parameter-efficient fine-tuning. The library leverages GPUs with Dask and RAPIDS to accelerate data curation, offering customizable and modular interfaces for pipeline expansion and model convergence. Key features include data download, text extraction, quality filtering, deduplication, downstream-task decontamination, distributed data classification, and PII redaction. NeMo Curator is suitable for curating high-quality datasets for large language model training.
dolma
Dolma is a dataset and toolkit for curating large datasets for (pre)-training ML models. The dataset consists of 3 trillion tokens from a diverse mix of web content, academic publications, code, books, and encyclopedic materials. The toolkit provides high-performance, portable, and extensible tools for processing, tagging, and deduplicating documents. Key features of the toolkit include built-in taggers, fast deduplication, and cloud support.
llm-datasets
LLM Datasets is a repository containing high-quality datasets, tools, and concepts for LLM fine-tuning. It provides datasets with characteristics like accuracy, diversity, and complexity to train large language models for various tasks. The repository includes datasets for general-purpose, math & logic, code, conversation & role-play, and agent & function calling domains. It also offers guidance on creating high-quality datasets through data deduplication, data quality assessment, data exploration, and data generation techniques.
For similar jobs
sweep
Sweep is an AI junior developer that turns bugs and feature requests into code changes. It automatically handles developer experience improvements like adding type hints and improving test coverage.
teams-ai
The Teams AI Library is a software development kit (SDK) that helps developers create bots that can interact with Teams and Microsoft 365 applications. It is built on top of the Bot Framework SDK and simplifies the process of developing bots that interact with Teams' artificial intelligence capabilities. The SDK is available for JavaScript/TypeScript, .NET, and Python.
ai-guide
This guide is dedicated to Large Language Models (LLMs) that you can run on your home computer. It assumes your PC is a lower-end, non-gaming setup.
classifai
Supercharge WordPress Content Workflows and Engagement with Artificial Intelligence. Tap into leading cloud-based services like OpenAI, Microsoft Azure AI, Google Gemini and IBM Watson to augment your WordPress-powered websites. Publish content faster while improving SEO performance and increasing audience engagement. ClassifAI integrates Artificial Intelligence and Machine Learning technologies to lighten your workload and eliminate tedious tasks, giving you more time to create original content that matters.
chatbot-ui
Chatbot UI is an open-source AI chat app that allows users to create and deploy their own AI chatbots. It is easy to use and can be customized to fit any need. Chatbot UI is perfect for businesses, developers, and anyone who wants to create a chatbot.
BricksLLM
BricksLLM is a cloud native AI gateway written in Go. Currently, it provides native support for OpenAI, Anthropic, Azure OpenAI and vLLM. BricksLLM aims to provide enterprise level infrastructure that can power any LLM production use cases. Here are some use cases for BricksLLM: * Set LLM usage limits for users on different pricing tiers * Track LLM usage on a per user and per organization basis * Block or redact requests containing PIIs * Improve LLM reliability with failovers, retries and caching * Distribute API keys with rate limits and cost limits for internal development/production use cases * Distribute API keys with rate limits and cost limits for students
uAgents
uAgents is a Python library developed by Fetch.ai that allows for the creation of autonomous AI agents. These agents can perform various tasks on a schedule or take action on various events. uAgents are easy to create and manage, and they are connected to a fast-growing network of other uAgents. They are also secure, with cryptographically secured messages and wallets.
griptape
Griptape is a modular Python framework for building AI-powered applications that securely connect to your enterprise data and APIs. It offers developers the ability to maintain control and flexibility at every step. Griptape's core components include Structures (Agents, Pipelines, and Workflows), Tasks, Tools, Memory (Conversation Memory, Task Memory, and Meta Memory), Drivers (Prompt and Embedding Drivers, Vector Store Drivers, Image Generation Drivers, Image Query Drivers, SQL Drivers, Web Scraper Drivers, and Conversation Memory Drivers), Engines (Query Engines, Extraction Engines, Summary Engines, Image Generation Engines, and Image Query Engines), and additional components (Rulesets, Loaders, Artifacts, Chunkers, and Tokenizers). Griptape enables developers to create AI-powered applications with ease and efficiency.