Data is the most valuable asset in LLM development. While datasets can't be directly evaluated like models, high-quality datasets have the following characteristics:

• Accuracy: Samples should be factually correct, helpful to users, and well-written. Answers should also be relevant to their corresponding instructions.
• Diversity: You want to cover as many use cases as possible to ensure proper instruction-following and relevant answers. This requires a wide range of topics, contexts, lengths, writing styles, etc. sampled in a representative way.
• Complexity: Answers should be nontrivial and a/ representative of tasks you expect the model to handle or b/ include complex tasks involving multi-step reasoning, planning, etc.

Measuring accuracy can be easy in the case of mathematical problems using a Python interpreter, or near-impossible with open-ended, subjective questions. On the other hand, clustering datasets by topic is a good way of measuring diversity. Finally, complexity can be assessed using other LLMs acting like judges.

Once a model has been pre-trained on a next-token prediction task, supervised fine-tuning is used to turn it into an assistant capable of answering questions and achieving tasks. These datasets contain pairs of instructions and outputs to train LLMs to go beyond their pre-training objective. All the datasets listed here should be under permissive licensing (Apache 2.0, MIT, cc-by-4.0, etc.).

The goal of general-purpose datasets is to transform base models into versatile and capable assistants by exposing them to a wide range of high-quality data. These datasets often include a diverse mix of real-world and synthetic data, commonly generated using models like GPT-4.

LLMs often struggle with mathematical reasoning and formal logic, which has led to the creation of specialized datasets. These datasets extend beyond pure mathematics, encompassing a wide range of problems that require systematic thinking and step-by-step reasoning, ultimately enabling LLMs to tackle complex real-world challenges that involve logical deduction and quantitative analysis.

Code is another challenging domain for LLMs that lack specialized pre-training. Code datasets, containing diverse programming language examples, are used to fine-tune LLMs and enhance their ability to understand, generate, and analyze code, enabling them to serve as effective coding assistants.

Many datasets focus on pairs of instructions and outputs, but chat models are often used in conversational settings. Conversational and role-play datasets expose LLMs to the patterns, nuances, and context-dependent nature of real conversations, allowing them to generate more natural, and engaging dialogues.

Function calling allows large language models (LLMs) to execute predefined functions with parameters inferred from user prompts, rather than generating standard text responses. This enables LLMs to seamlessly integrate with external systems, perform complex operations, and provide more accurate and contextually relevant responses.

W.I.P.

To create a high-quality dataset, focus on carefully curating a diverse set of relevant, accurate, and informative examples rather than simply maximizing dataset size.

Start by aggregating available data from various sources (open-source or not) and applying filters like data deduplication and data quality. If the initial dataset is small or insufficient, consider synthetically generating additional data that mirrors its quality and style. Iteratively explore and refine the dataset by assessing model performance, identifying gaps, and collecting or generating data to address those shortcomings.

Tools listed in this section may belong to several categories but appear in only one for clarity.

• Exact deduplication: Remove identical samples with data normalization (e.g., convert text to lowercase), hash generation (e.g., create an MD5 or SHA-256 hash for each sample), and duplicate removal.
• Fuzzy deduplication
  • MinHash: Fuzzy deduplication with hashing, sorting, and Jaccard similarity (preferred technique).
  • BLOOM filters: Fuzzy deduplication with hashing and fixed-size vector.
• Decontamination: Remove samples too close to test sets, using either exact or fuzzy filtering.

• Rule-based filtering: Remove samples based on a list of unwanted words, like refusals and "As an AI assistant" (example).
• Argilla: Open-source data curation platform that allows you to filter and annotate datasets in a collaborative way.
• LLM-as-a-judge: Colab notebook that provides code to rate outputs with Mixtral-7x8B.
• Data Prep Kit: Framework for data preparation for both code and language, with modules in Python, Ray, and Spark, and a wide range of scale from laptops to data centers.
• DataTrove: HuggingFace library for large-scale data processing, used in the creation of Fineweb.

• Distilabel: General-purpose framework that can generate and augment data (SFT, DPO) with techniques like UltraFeedback and DEITA.
• Auto Data: Lightweight library to automatically generate fine-tuning datasets with API models.
• Bonito: Library for generating synthetic instruction tuning datasets for your data without GPT (see also AutoBonito).
• Augmentoolkit: Framework to convert raw text into datasets using open-source and closed-source models.
• Magpie: Your efficient and high-quality synthetic data generation pipeline by prompting aligned LLMs with nothing.
• Genstruct: An instruction generation model, which is designed to generate valid instructions from raw data.
• DataDreamer: A python library for prompting and synthetic data generation.

• llm-swarm: Generate synthetic datasets for pretraining or fine-tuning using either local LLMs or Inference Endpoints on the Hugging Face Hub.
• Cosmopedia: Hugging Face's code for generating the Cosmopedia dataset.
• textbook_quality: A repository for generating textbook-quality data, mimicking the approach of the Microsoft's Phi models.

• sentence-transformers: A python module for working with popular language embedding models.
• Lilac: Tool to curate better data for LLMs, used by NousResearch, databricks, cohere, Alignment Lab AI. It can also apply filters.
• Nomic Atlas: Interact with instructed data to find insights and store embeddings.
• text-clustering: A framework from Huggingface for clustering textual data.
• BunkaTopics: Data cleaning and topic modeling visualization.
• Autolabel: Automatically label data using popular language models.

• Trafilatura: Python and command-line tool to gather text and metadata on the web. Used for the creation of RefinedWeb.
• Marker: Quickly convert PDFs to markdown text.

Special thanks to geronimi73, Bytes-Explorer, and euclaise for their PRs.

Please let me know if a dataset is not properly credited.

• Wei-Lin Chiang et al, "Vicuna: An Open-Source Chatbot Impressing GPT-4 with 90%* ChatGPT Quality," 2023.
• Yihan Cao et al, "Instruction Mining: When Data Mining Meets Large Language Model Finetuning," 2023.
• Subhabrata Mukherjee et al, "Orca: Progressive Learning from Complex Explanation Traces of GPT-4," 2023.
• Chunting Zhou et al, "LIMA: Less Is More for Alignment," 2023.
• Suriya Gunasekar et al, "Textbooks Are All You Need," 2023.
• Lichang Chen et al, "AlpaGasus: Training A Better Alpaca with Fewer Data," 2024.
• Zheng Cai et al, "InternLM2 Technical Report," 2024.
• Lifan Yuan et al, "Advancing LLM Reasoning Generalists with Preference Trees," 2024.
• Wei Liu et al, "What Makes Good Data for Alignment? A Comprehensive Study of Automatic Data Selection in Instruction Tuning," 2024.
• Xingyao Wang et al, "MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback," 2024.