Zynerji/Ektome-granite-3.0-8b-instruct-PristinelyUncensored
Zynerji's Ektome-granite-3.0-8b-instruct-PristinelyUncensored is an 8 billion parameter instruction-tuned language model with a 32K context length. This model is uncensored by design, utilizing the proprietary Ektomē method to remove refusal-specific components while preserving general helpfulness and knowledge. It is intended for use cases where an uncensored model is explicitly desired, though no capability retention certificate has been run for this specific version.
Loading preview...
Ektome-granite-3.0-8b-instruct-PristinelyUncensored Overview
This model, developed by Zynerji, is an 8 billion parameter instruction-tuned language model with a 32K context length. It is notable for being uncensored by construction, meaning it will not refuse prompts. The model employs a proprietary method called Ektomē (excision) to isolate and remove only the refusal-specific components from the base model, aiming to leave general helpfulness and reasoning capabilities intact without additional training or distillation.
Key Characteristics
- Uncensored Design: Explicitly designed not to refuse prompts, achieved through the Ektomē method.
- Proprietary Excision: Uses a unique, norm-preserving process to remove refusal mechanisms without damaging core capabilities.
- Quantizations Available: Provided in various GGUF quantizations, including
Q8_0,Q6_K,Q5_K_M,Q4_K_M,IQ4_XS, andIQ3_M, withIQ*variants utilizing imatrix quantization for better quality at lower precision.
Important Considerations
- No Certification: This specific model version has not undergone the n=2800 paired certificate test for capability retention. Any reported numbers are point estimates without confidence intervals.
- Limitations: The Ektomē certificate, when run, only bounds capability retention and does not certify safety, factual accuracy, or fitness for any purpose. Users are accountable for the outputs generated by this uncensored model.