Hi r\/datasets<\/a> community!<\/p>\n I’m the creator of DeepFabric<\/strong> (https:\/\/github.com\/lukehinds\/deepfabric<\/a>), an open-source tool that generates synthetic datasets using LLMs and novel approaches leveraging graphs (DAG) and Trees. I’m looking for collaborators who need custom datasets and are willing to provide feedback on quality and usefulness.<\/p>\n What DeepFabric does:<\/strong> DeepFabric creates diverse, domain-specific synthetic datasets using a unique graph\/tree-based architecture. It generates data in OpenAI chat format with more formats coming, minimizes redundancy through structured topic generation.<\/p>\n What I’m offering:<\/strong> I’ll create custom synthetic datasets tailored to your specific domain or use case, cover all LLM API costs myself, provide technical support and customization, and generate datasets ranging from small proof-of-concepts to larger training sets.<\/p>\n What I’m looking for:<\/strong> I need detailed feedback on dataset quality, diversity, and usefulness, insights into how well the synthetic data performs for your specific use case, suggestions for improvements or missing features, and optionally a brief case study write-up of your experience.<\/p>\n Ideal collaborators:<\/strong> I’m particularly interested in working with researchers or developers working in a professional capacity, doing model distillation or evaluation benchmarks, or anyone needing training data for specialized or niche domains for machine learning \/ statistical analysis – a good example might be people working with limited real-world data availability. I have so far received really good feedback from a medical professor who needed data around mock scenarios of someone complaining about symptoms that could signal risk of heart attack.<\/p>\n Examples of what I can generate:<\/strong> Think Q&A pairs for specific technical domains, conversational data for chatbot training, domain-specific instruction-following datasets, or evaluation benchmarks for specialized tasks. I am also able to convert to whatever format you need.<\/p>\n If you’re interested, please comment or PM with your domain\/use case, approximate dataset size needed, brief description of your intended use, and timeline if you have one.<\/p>\n I’ll prioritize collaborations that offer the most learning opportunities for both of us. Looking forward to working with some of you!<\/p>\n Some examples: medical Q&A: https:\/\/huggingface.co\/datasets\/lukehinds\/medical_q_and_a<\/a><\/p>\n Programming Challenges: https:\/\/huggingface.co\/datasets\/lukehinds\/programming-challenges-one<\/a><\/p>\n Repository:<\/strong> https:\/\/github.com\/lukehinds\/deepfabric<\/a> submitted by \/u\/DecodeBytes <\/a>
Documentation:<\/strong> https:\/\/lukehinds.github.io\/DeepFabric\/synethic<\/a> data<\/p>\n<\/div>\n
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
