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

    Neuronpedia

    Jacobian LensNEW
    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. Llama3.1-8B-IT
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 38028
    Prev
    Next
    INDEX
    Explanations

    Violence, debt, or crime

    np_max-act · gemini-2.0-flash

    This neuron responds most strongly to tokens that mark descriptive, scene-setting narrative passages—i.e. when the text shifts into describing the environment, characters’ positions, and their actions.

    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
    ushort
    -0.07
    ramid
    -0.07
    bec
    -0.07
    +</
    -0.06
    xb
    -0.06
     Predicate
    -0.06
     }:
    -0.06
    Wizard
    -0.06
     plá
    -0.06
    Climate
    -0.06
    POSITIVE LOGITS
    rieg
    0.07
    Sugar
    0.06
     Socket
    0.06
    URRE
    0.06
    ITT
    0.06
    ecc
    0.06
    ellaneous
    0.05
    ?)
    0.05
    ności
    0.05
     —
    0.05
    Activations Density 0.008%

    No Known Activations