mlabonne / llm-datasets

High-quality datasets, tools, and concepts for LLM fine-tuning.

🐦 Follow me on X • 🤗 Hugging Face • 💻 Blog • 📙 Hands-on GNN

Hi, I'm a Machine Learning Scientist, Author, Blogger, and LLM Developer.

💼 Projects

• The LLM Course: A popular curated list of resources to get into LLMs (>29k ⭐).
• Hands-on GNNs: My book about graph neural networks published by Packt (all the code is open source).
• LLM Datasets: Curated list of high-quality datasets for LLM fine-tuning.
• LLM Tools: Automate LLM pipelines with Colab notebooks like LLM AutoEval, LazyMergekit, LazyAxolotl, and AutoQuant.

🤗 Models

• AlphaMonarch-7B: Top performer in terms of reasoning + conversational abilities on a variety of benchmarks. [Demo]
• NeuralBeagle14-7B: The most powerful 7B model (rank 10 on the entire Open LLM Leaderboard). [Demo]
• Phixtral: Novel Mixture of Experts architecture with phi-2 models. [Demo]
• Beyonder-4x7B-v3: Mixture of Experts with four excellent fine-tuned Mistral-7b models. [Demo]
• NeuralMarcoro14: My previous best 7B model (rank 1 on the Open LLM Leaderboard 7B param). [Demo]
• NeuralHermes: A DPO fine-tuned version of OpenHermes (extremely cost-efficient). [Demo]

llm-datasets's Issues

How to create an instruction dataset from .pdf and .docx documents

Hello I'm in the process of fine-tuning a Large Language Model (LLM) for an NGO and I need to construct an instruction dataset from .pdf and .docx documents containing information in text.

The objective is to extract instructions from these documents and organize them into a structured dataset suitable for fine-tuning the LLM. This involves parsing .pdf and .docx files, extracting relevant text segments, and annotating them.

I'm seeking guidance and recommendations from the community on how to efficiently create this dataset. Specifically, I'm interested in:

1. Techniques and libraries for parsing .pdf and .docx documents in Python.
2. Strategies for extracting instructional content from the parsed documents while maintaining context and fidelity.
3. Approaches for annotating the extracted text segments as instructional content, including identifying key actions, steps, and contextual information.

Any advice, best practices, or resources you can provide to assist in this endeavor would be greatly appreciated. Thank you for your support!