Green-Eye/Llama-3.2-1B-Instruct-heretic

TEXT GENERATIONPricing:Input $0.108 / Output $0.804Concurrent Unit Cost:1Model Size:1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 30, 2026License:llama3.2Architecture:Transformer Featherless Exclusive Cold

Green-Eye/Llama-3.2-1B-Instruct-heretic is a 1 billion parameter instruction-tuned causal language model developed by RACER IS OP, based on Meta's Llama-3.2 architecture with a 32768 token context length. This variant is decensored using Heretic v1.2.0's directional ablation, suppressing refusal behavior via targeted weight edits rather than fine-tuning. It is optimized for use cases requiring an uncensored model, such as local agents, roleplay, or research, while retaining the base model's knowledge and instruction-following capabilities.

Loading preview...

Green-Eye/Llama-3.2-1B-Instruct-heretic Overview

This model is a 1 billion parameter instruction-tuned variant of Meta's Llama-3.2-1B-Instruct, developed by RACER IS OP. It has been modified using the Heretic v1.2.0 "abliteration" technique, which involves targeted weight edits to the attention output and MLP down-projections. This process effectively suppresses refusal behavior without extensive fine-tuning, largely preserving the base model's original knowledge and instruction-following abilities.

Key Characteristics

  • Decensored Behavior: Significantly reduces refusal rates (from 96/100 to 7/100 adversarial prompts) compared to the original Llama-3.2-1B-Instruct.
  • Targeted Modification: Achieves decensoring with a low KL divergence (0.1713), indicating a narrow and precise edit that minimizes impact on other capabilities.
  • Lightweight: At 1 billion parameters, it is suitable for on-device deployment, mobile, or edge computing scenarios.
  • Broad Compatibility: Provided with a full suite of GGUF quantizations, enabling efficient execution on various hardware, including gaming PCs, with specific recommendations for GPU memory footprints.

Ideal Use Cases

  • Local Agents & Roleplay: Provides an uncensored foundation for applications where creative freedom and non-refusal are critical.
  • Alignment Research: Useful for studying refusal mechanics and the effects of alignment techniques.
  • Resource-Constrained Environments: Its small size and GGUF support make it excellent for deployment on devices with limited computational resources.

Finetunes
1 models