kmseong/llama2_7b_chat_only_sn_tuned_lr3e-5

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Apr 17, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The kmseong/llama2_7b_chat_only_sn_tuned_lr3e-5 is a 7 billion parameter Llama 2-based chat model, specifically a Safety Neuron-Tuned (SN-Tune) version of Llama-3.2-3B-Instruct. It has been fine-tuned using the SN-Tune method on safety alignment data to enhance safety while preserving general capabilities. This model is optimized for applications requiring improved safety alignment and parameter-efficient fine-tuning.

Loading preview...

Model Overview

This model, kmseong/llama2_7b_chat_only_sn_tuned_lr3e-5, is a 7 billion parameter chat-optimized variant based on the Llama 2 architecture. It is a Safety Neuron-Tuned (SN-Tune) version of meta-llama/Llama-3.2-3B-Instruct, developed by kmseong.

Key Capabilities & Features

  • Safety Neuron Tuning (SN-Tune): Employs a unique fine-tuning approach where only a small set of "safety neurons" are fine-tuned on safety data, while other parameters remain frozen. This method is designed to:
    • Enhance safety alignment significantly.
    • Minimize impact on the model's general capabilities and performance.
    • Provide a parameter-efficient fine-tuning solution.
  • Base Model: Built upon meta-llama/Llama-3.2-3B-Instruct, inheriting its foundational language understanding and generation capabilities.
  • Context Length: Supports a context length of 4096 tokens.

Use Cases & Strengths

This model is particularly well-suited for applications where enhanced safety alignment is a critical requirement, without sacrificing the broad utility of a general-purpose chat model. Its SN-Tune methodology makes it an efficient choice for developers looking to deploy safer AI interactions with minimal computational overhead for fine-tuning. It is ideal for chat applications, customer support, and content moderation where mitigating harmful outputs is paramount.