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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Qwen2.5-7B-IT
    4. Resid Post - 131k
    5. 15-RESID-POST-AA
    6. 52765
    Prev
    Next
    INDEX
    Explanations

    Detects first-person self-references (the token "I"/"I'm") in assistant replies, especially in apologetic or refusal statements.

    oai_token-act-pair · gpt-5-miniTriggered 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
     sad
    -0.07
    奋
    -0.07
     Ba
    -0.06
     supervised
    -0.06
    bar
    -0.06
    颛
    -0.06
    相
    -0.06
     Cool
    -0.06
    AGE
    -0.06
     Vì
    -0.06
    POSITIVE LOGITS
     haci
    0.08
    决议
    0.08
    的位置
    0.08
    .minecraftforge
    0.08
     �
    0.07
     citrus
    0.07
     지역
    0.07
    _closure
    0.07
    allenge
    0.07
    文件
    0.07
    Activations Density 0.042%

    No Known Activations