kmseong/llama3.2_3b_SSFT_epoch5_adam
TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:Apr 3, 2026License:llama3.2Architecture:Transformer Loading

The kmseong/llama3.2_3b_SSFT_epoch5_adam is a 3.2 billion parameter Llama 3.2-based causal language model developed by Min-Seong Kim. This model has undergone Phase 0 of Safety-WaRP (Weight space Rotation Process) training, specifically focusing on base safety training using the Circuit Breakers dataset. It is designed to establish fundamental safety mechanisms, making it suitable as a foundational model for further safety and utility enhancements.

Loading preview...