© 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
    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. 62029
    Prev
    Next
    INDEX
    Explanations

    This neuron detects user instructions or directives addressed to the assistant—imperative/request phrases (including polite markers like "please" and verbs telling the assistant what to do).

    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
     complains
    -0.08
     Tic
    -0.07
    묀
    -0.07
    mówi
    -0.07
     мил
    -0.07
     communicating
    -0.07
    دل
    -0.07
    맴
    -0.06
     trivial
    -0.06
    是真的
    -0.06
    POSITIVE LOGITS
    little
    0.07
    (from
    0.06
    _gradient
    0.06
    尺
    0.06
    ᐖ
    0.06
    ished
    0.06
    ⽬
    0.06
    ,X
    0.06
    <!--
    0.06
    .with
    0.06
    Activations Density 0.136%

    No Known Activations