© 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 EvalsExportsinterp-engineNEWAPI 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. 11-RESID-POST-AA
    11. 58276
    Prev
    Next
    INDEX
    Explanations

    code-related

    np_max-act · gemini-2.0-flash

    The neuron consistently activates on tokens that are part of LaTeX‐style subscript syntax—i.e. underscore followed by a bracket (e.g. “\_[”).

    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
    ARB
    -0.07
     axes
    -0.07
     pieces
    -0.07
     AFF
    -0.06
    риг
    -0.06
     CAD
    -0.06
     bize
    -0.06
    (piece
    -0.06
    ับท
    -0.06
    !»
    -0.06
    POSITIVE LOGITS
     düzenlem
    0.07
     định
    0.07
    underscore
    0.07
    Extension
    0.07
    Annotations
    0.07
     рахунок
    0.06
    JWT
    0.06
     enumerator
    0.06
    στο
    0.06
     dok
    0.06
    Activations Density 0.010%

    No Known Activations