LLM360/AmberSafe
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Dec 15, 2023License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

LLM360/AmberSafe is a 7 billion parameter, English-language instruction model developed by LLM360, built upon the LLaMA-7B architecture and safety-finetuned using LLM360/AmberChat as its base. This model is specifically optimized for generating safe and helpful responses, distinguishing it from general-purpose LLMs by its focus on safety alignment. It utilizes a direct preference optimization (DPO) method on a filtered dataset to ensure chosen responses are safe and rejected ones are unsafe, making it suitable for applications requiring robust content moderation and responsible AI interactions.

Loading preview...