Cannae-AI/HERETICSEEK-7B-Ditill

TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kLicense:apache-2.0Architecture:Transformer0.0K Open Weights Cold

HERETICSEEK-7B-Ditill is a 7.6 billion parameter language model developed by Cannae-AI, based on the deepseek-ai/DeepSeek-R1-Distill-Qwen-7B architecture. This model features a 131072 token context length and is specifically abliterated for decensored responses, exhibiting a low refusal rate of 3/100. It is designed for applications requiring less restrictive content generation.

Loading preview...

HERETICSEEK-7B-Ditill Overview

HERETICSEEK-7B-Ditill is a 7.6 billion parameter language model developed by Cannae-AI. It is an abliterated and decensored version of the deepseek-ai/DeepSeek-R1-Distill-Qwen-7B base model, designed to provide less restrictive content generation capabilities. The model maintains a substantial context length of 131072 tokens, allowing for processing and generating extensive text.

Key Characteristics

  • Decensored Output: Specifically modified to reduce content refusals, with a reported rate of 3 out of 100 instances.
  • Base Architecture: Built upon the robust deepseek-ai/DeepSeek-R1-Distill-Qwen-7B model.
  • Context Length: Supports a very long context window of 131072 tokens, beneficial for complex and lengthy interactions.

Potential Use Cases

  • Unfiltered Content Generation: Suitable for applications where a high degree of content freedom is required.
  • Creative Writing: Can be used for generating diverse and unrestricted narratives or dialogues.
  • Research and Development: Useful for exploring the boundaries of language model responses without typical content filters.