© 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. Llama3.1-8B-IT
    7. Resid Post - 131k
    9. 15-RESID-POST-AA
    11. 29931
    Prev
    Next
    INDEX
    Explanations

    requests or instructions

    np_max-act · gemini-2.0-flash

    This neuron detects polite, helpful assistant-style wording — first-person apologies, offers, and reassuring/supportive phrases (I/we statements offering help or accommodation).

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

    warm, empathetic customer‑service style responses that reassure, apologize, and offer helpful alternatives in an email or support context.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-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
     δη
    -0.07
    <Tag
    -0.07
     sáng
    -0.06
     กระ
    -0.06
     武
    -0.06
     arsen
    -0.06
    ffa
    -0.06
     luyện
    -0.06
    .department
    -0.06
    hibited
    -0.06
    POSITIVE LOGITS
     "./
    0.07
    İS
    0.06
    उ
    0.06
    EDITOR
    0.06
     Zuckerberg
    0.06
    Spoiler
    0.06
    ㅠㅠ
    0.06
     regrets
    0.06
    _BEFORE
    0.06
    RODUCTION
    0.06
    Activations Density 0.295%

    No Known Activations