Automorphic introduces Conduit, an innovative tool designed to empower language models with enriched knowledge. Conduit surpasses the limitations of traditional prompt stuffing by enabling fine-tuning of language models, training adapters for specific behaviors or knowledge domains, and seamlessly combining them in dynamic ways.

Key Features:

1. Streamlined Model Enhancement: Conduit accelerates model development through iterative refinement, incorporating human-in-the-loop feedback, and simplifying model deployment.
2. Continuous Model Improvement: Conduit ensures language models evolve and adapt to user needs by incorporating real-time feedback and manual labeling.
3. Efficient Adapter Integration: Achieve high-performance and deployment efficiency with Conduit’s ability to quickly load and integrate fine-tuned adapters. Compatibility with the OpenAI API allows seamless integration into existing codebases.
4. Automorphic Hub: Explore the Automorphic Hub, a platform hosting publicly shared models trained and enriched with Automorphic’s capabilities. Access these models to enhance your own applications and projects.

Automorphic also offers TREX, a versatile tool that converts unstructured data into a structured format of your choice, including JSON, XML, YAML, or custom formats defined by regular expressions or context-free grammars. TREX provides a highly customizable alternative to OpenAI’s functions.

Additionally, Aegis, another offering from Automorphic, acts as a formidable firewall tool, safeguarding language models and users from adversarial attacks. Aegis defends against various threats, including prompt injections, personal identifiable information (PII) leaks, and toxic language. It continually learns from usage patterns to enhance its detection capabilities over time, ensuring robust protection.