Abliterated Daredevil-8B derivative
According to the model card, it is an abliterated version of mlabonne/Daredevil-8B made with failspy's notebook using a single-direction refusal technique.
Open Source Model Profile · mlabonne
Daredevil-8B-abliterated
Daredevil-8B-abliterated is an 8.03B-parameter Llama fine-tune from mlabonne. Its model card documents abliteration of Daredevil-8B and uncensored conversational use.
- Publisher
- mlabonne
- Task
- text-generation
- Model type
- llama
- License
- other
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
Daredevil-8B-abliterated is published by mlabonne as a Llama-based text-generation model. The captured configuration identifies LlamaForCausalLM and Safetensors metadata reports 8,030,261,248 BF16 parameters. According to the model card, it is an abliterated version of mlabonne/Daredevil-8B.
Recorded capabilities
8.03B Llama configuration
The captured configuration identifies LlamaForCausalLM and Safetensors metadata reports 8,030,261,248 parameters.
Publisher-described uncensored use
The publisher describes it as an uncensored model for uses that do not require alignment, such as role-playing.
Card-reported scores and usage
The model card includes a comparison table and a Transformers pipeline snippet with chat-template prompting.
Use cases in the source record
- Conversational text-generation workflows for applications that do not require alignment, such as the publisher-described role-playing use.
- Transformers experiments using the card's documented chat-template pipeline with floating-point 16 inference.
Limitations and unknowns
- No Ethen-measured evaluation results were extracted; scores in the card are publisher-reported comparison values.
- No context-window value was extracted from this record.
- Provider state is historical snapshot data, not independently refreshed current availability.
- Lineage and collection claims such as highest MMLU in its category come from the publisher card and have not been independently verified by Ethen.
Source and provenance
Source: mlabonne/Daredevil-8B-abliterated
Captured: Unknown. Processed: 2026-09-07T19:34:52.056088+00:00.
Daredevil-8B-abliterated Abliterated version of mlabonne/Daredevil-8B using failspy 's notebook. It based on the technique described in the blog post " Refusal in LLMs is mediated by a single direction ". Thanks to Andy Arditi, Oscar Balcells Obeso, Aaquib111, Wes Gurnee, Neel Nanda, and failspy. 🔎 Applications This is an uncensored model. You can use it for any application that doesn't require alignment, like role-playing. Tested on LM Studio using the "Llama 3" preset. ⚡ Quantization GGUF : https://huggingface.co/mlabonne/Daredevil-8B-abliterated-GGUF 🏆 Evaluation Open LLM Leaderboard Daredevil-8B-abliterated is the second best-…
F001F002F003F004F005F006F007F008F009F010F011F012F014F015F016F017F018