© 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. 11-RESID-POST-AA
    11. 7780
    Prev
    Next
    INDEX
    Explanations

    block

    np_max-act · gemini-2.0-flash

    This neuron responds to the instruction‐style phrase “what to do and what not to do,” specifically picking up on the “to do … and … not to do” pattern in directive text.

    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
    _rating
    -0.07
    Node
    -0.07
     tweet
    -0.07
     Paths
    -0.07
     slightest
    -0.07
     прев
    -0.06
    .Compare
    -0.06
    학회
    -0.06
    postId
    -0.06
    .SimpleButton
    -0.06
    POSITIVE LOGITS
     дер
    0.06
     referee
    0.06
    elle
    0.06
    .'</
    0.05
     Auto
    0.05
     Cobra
    0.05
     espec
    0.05
     ἐ
    0.05
     трех
    0.05
    attro
    0.05
    Activations Density 0.000%

    No Known Activations