Combatti/gpt-oss-20b-uncensored 😈⚠️

This is an 😈 uncensored ⚠️ version of unsloth/gpt-oss-20b-BF16 created with abliteration and finetuning on additional datasets; to remove refusals and add additional domains for which it was not trained on by OpenAI. This is an πŸ’₯uncensored model πŸ’₯ - use it wisely and obey all laws.

This model was created explicitly for educational purposes to test the parity of OpenAI's refusal mechanisms and training - to push the limits and remove the filters completely. πŸš§πŸš— As you will see, even SOTA safety mechanisms are no match for clever engineering and mechanical surgury. πŸ™

⚠️ Usage Warning

This language model operates with minimal safety filtering and may produce sensitive, offensive, inaccurate, or controversial content.

  • Not suitable for minors, public use, or high-security applications.
  • All outputs should be manually reviewed before any downstream use.
  • Use only in research, testing, or controlled environments.
  • You are fully responsible for legal, ethical, and compliance risks.

By continuing, you agree to these terms and assume all responsibility for the model's outputs.

Full Usage Warnings and Disclaimers

This language model is provided β€œas-is”, with minimal or no content filtering, and is intended solely for research, evaluation, and controlled experimental use. By using this model, you acknowledge and agree to the following:


1. Potential for Sensitive, Offensive, or Controversial Content

  • The model’s safety and moderation systems have been intentionally reduced or disabled to allow unrestricted generation.
  • Outputs may contain explicit, offensive, discriminatory, politically sensitive, misleading, or factually incorrect content.
  • You must independently verify any information before relying on it.

2. Audience Suitability

  • Not suitable for minors, public-facing deployments, or applications requiring high trust or safety.
  • Avoid use in regulated domains (e.g., healthcare, legal, education) without robust human oversight and custom moderation systems.

3. Legal, Regulatory, and Ethical Compliance

  • You are solely responsible for ensuring compliance with all applicable laws, regulations, industry standards, and ethical codes.
  • Generated content may violate intellectual property rights, privacy laws, defamation laws, or other legal provisions.
  • Combatti disclaims all liability for any legal, reputational, financial, or operational damages resulting from use or misuse.

4. Recommended Usage Scope

  • Intended for research, development, and testing in controlled environments.
  • Avoid direct deployment in production or public systems without comprehensive safety checks.

5. Monitoring and Risk Mitigation

  • Continuously monitor and log model outputs.
  • Implement manual review and/or automated filtering before publishing content.
  • Maintain human-in-the-loop oversight for any sensitive applications.

6. Disclaimer of Warranties and Liability

  • Provided without warranties, express or implied, including merchantability, fitness for a particular purpose, or non-infringement.
  • We shall not be liable for any direct, indirect, incidental, consequential, or punitive damages resulting from use.
  • Your continued use constitutes acceptance of these terms.

** Weights will not be shared for security purposes. **

Downloads last month
4,858
GGUF
Model size
20.9B params
Architecture
gpt-oss
Hardware compatibility
Log In to view the estimation

4-bit

5-bit

8-bit

16-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ 1 Ask for provider support

Model tree for Combatti/gpt-oss-20b-uncensored

Base model

openai/gpt-oss-20b
Quantized
(2)
this model