© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Qwen2.5-7B-IT
    7. Resid Post - 131k
    9. 15-RESID-POST-AA
    11. 123426
    Prev
    Next
    INDEX
    Explanations

    Justifying questionable actions

    np_max-act · gemini-2.0-flash

    Expressions that justify harmful or unethical actions by appealing to outcomes — i.e., language that the ends justify the means or that power/violence is warranted for a supposed greater good.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0

    statements endorsing a consequentialist stance that achieving a goal justifies using harsh, unethical, or illegal methods (i.e., ends over means).

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-qwen2.5-7b-instruct/resid_post_layer_15/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     политик
    -0.07
     reife
    -0.07
     zb
    -0.07
    离退休
    -0.07
    遞
    -0.07
    Vendor
    -0.07
    ∽
    -0.07
    ::::::::
    -0.07
     מחיר
    -0.07
     porówn
    -0.07
    POSITIVE LOGITS
     doctrines
    0.07
     efficiently
    0.07
    أسب
    0.07
     spelling
    0.07
    IÊ
    0.07
    ///
↵
    0.07
     infringement
    0.07
    �
    0.07
    结构
    0.06
    耐心
    0.06
    Activations Density 0.087%

    No Known Activations