Hum, the leading provider of AI & data intelligence solutions for publishers, today announced the open source release of their new large language model (LLM), Lodestone.
Lodestone is the new LLM model powering Alchemist. It creates rich, AI-native representations called embeddings which serve as the foundation for understanding content, topics, people, and organizations. For this reason it's often called a “foundation model”. Optimized for long text sequences, Lodestone can read a mid-length research paper. Similar models from Google, Microsoft, Facebook, and LLM researchers only read the first one or two paragraphs.
Lodestone is the highest performing model of its size and sequence length on the MTEB Leaderboard.
With its newfound powers, Alchemist improves its ability to:
- Find richer audiences likely to be interested in a topic
- Derive deep content insights and understanding
- Power hyper-personalized content recommendations
- Derive better topics as part of the Alchemist Tagging Engine
- Power enhanced content deep search
Lodestone was trained on a large publicly available dataset, including over 1 million scholarly research articles and publications. The model can process text sequences of 4096 tokens to better capture topical context and nuance compared to other commonly-used LLM models.
“We are building general AI capabilities for the publisher community, and we’re glad to share along the way,” said Dustin Smith, Hum’s Co-Founder & President. “Lodestone represents an upgraded foundation for Alchemist, and can be adapted for a wide array of AI applications.”
Starting today, developers and enterprises can fine-tune and deploy their own models using Lodestone on Hugging Face, putting long-sequence AI applications in reach of more projects and publishers.
"This release furthers our commitment to using AI to solve challenges for publishers and societies," said Niall Little, Hum’s CTO. "We’re continuing to fine-tune the model for the needs of our publishing industry clients, but look forward to seeing how the community can build on Lodestone to advance applications for content intelligence and responsible AI."