2,901 1 week ago

uncensored gpt-oss

tools thinking 20b 120b
ollama run gurubot/gpt-oss-derestricted:120b

Details

1 week ago

60b935956ce3 · 88GB ·

gpt-oss
·
117B
·
Q4_K_M
<|start|>system<|message|>You are ChatGPT, a large language model trained by OpenAI. Knowledge cutof
Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR US
{ "temperature": 1 }

Readme

I fixed the templates and merged the split GGUFs (for the 120b) both are Q4_K_M image.png

removed refusal behaviors while maintaining the high-performance reasoning of the original gpt-oss-20b. This is unlike regular abliteration which often inadvertently "lobotomizes" the model.

Methodology: Norm-Preserving Biprojected Abliteration
To achieve this, Arli AI utilized Norm-Preserving Biprojected Abliteration, a refined technique pioneered by Jim Lai (grimjim).

GGUF source: https://huggingface.co/mradermacher/gpt-oss-20b-Derestricted-GGUF

https://huggingface.co/mradermacher/gpt-oss-120b-Derestricted-GGUF

Original model source: https://huggingface.co/ArliAI/gpt-oss-20b-Derestricted

https://huggingface.co/ArliAI/gpt-oss-120b-Derestricted