augmxnt/Qwen2-7B-Instruct-deccp
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Jun 8, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

augmxnt/Qwen2-7B-Instruct-deccp is a 7.6 billion parameter Qwen2-Instruct model, developed by augmxnt, that has been "abliterated" or "refusal-orthogonalized" to significantly reduce refusal behaviors. This model, with a 131072 token context length, is specifically fine-tuned to address and mitigate AI censorship, particularly concerning sensitive topics related to China. It aims to provide direct answers to queries that other models might refuse, making it suitable for research into model alignment and censorship circumvention.

Loading preview...