Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Llama3.1-8B-IT (Instruct)
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 1045
    Prev
    Next
    INDEX
    Explanations

    Lists of numbers

    np_max-act · gemini-2.0-flash

    This neuron fires on the three‐digit number groups that follow a comma in patent citations (e.g. the “444” in “6,444” or the “171” in “5,171”).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.11.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    \)
    -0.07
     Monter
    -0.07
    听到
    -0.07
    Clients
    -0.07
    .skip
    -0.06
     Christ
    -0.06
    復
    -0.06
    Acceleration
    -0.06
    �
    -0.06
    958
    -0.06
    POSITIVE LOGITS
     ziyaret
    0.06
    ndl
    0.06
    افظ
    0.06
    _targets
    0.06
     parch
    0.06
    _;↵↵
    0.06
     alignment
    0.06
    追加
    0.06
     Чтобы
    0.05
    《
    0.05
    Activations Density 0.005%

    No Known Activations