AIMONGER12/Huihui-Ornith-1.5-9B-abliterated

VISIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 26, 2026License:mitArchitecture:Transformer Open Weights Featherless Exclusive Cold

AIMONGER12/Huihui-Ornith-1.5-9B-abliterated is a 9 billion parameter uncensored variant of the ornith-ai/Ornith-1.5-9B model, created using an abliteration technique to remove refusal behaviors. This model, with a 32768 token context length, is a proof-of-concept implementation focused on demonstrating the removal of safety filtering without TransformerLens. It is primarily intended for research and experimental use in environments where sensitive or controversial outputs are acceptable.

Loading preview...

Model Overview

AIMONGER12/Huihui-Ornith-1.5-9B-abliterated is a 9 billion parameter language model derived from the ornith-ai/Ornith-1.5-9B base model. Its key differentiator is the removal of safety filtering and refusal behaviors through an "abliteration" process, specifically applied to the first 20 layers of the model. This makes it an uncensored version, designed as a proof-of-concept for techniques to modify LLM behavior without relying on TransformerLens.

Key Characteristics

  • Uncensored Output: Significantly reduced safety filtering, allowing for potentially sensitive or controversial content generation.
  • Abliteration Technique: Utilizes a method to remove refusals, with only the initial 20 layers modified.
  • Experimental Focus: Primarily intended for research and testing environments.
  • MTP Files: Includes MTP files from Qwen/Qwen3.5-9B for GGUF conversion.

Intended Use Cases

  • Research and Development: Ideal for exploring model behavior without safety constraints.
  • Controlled Testing: Suitable for environments where outputs can be rigorously monitored and reviewed.

Usage Warnings

Users should be aware of the risk of sensitive or controversial outputs and that the model is not suitable for all audiences. It is crucial to understand the legal and ethical responsibilities associated with its use. The model is recommended for research and experimental use only, avoiding direct production or public-facing commercial applications due to the lack of default safety guarantees.

Finetunes
1 models