© 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 (Instruct)
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 43436
    Prev
    Next
    INDEX
    Explanations

    problems or conflicts

    np_max-act · gemini-2.0-flash

    This neuron activates on modal and conditional auxiliary verbs (e.g., can, would, shouldn’t), marking clauses that express possibility, necessity, or hypothetical situations.

    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
    esar
    -0.06
    -visible
    -0.06
     arguing
    -0.06
    (par
    -0.06
    anske
    -0.06
    Suc
    -0.06
     가진
    -0.06
    zeichnet
    -0.06
     //</
    -0.06
    करण
    -0.06
    POSITIVE LOGITS
    _scan
    0.07
    xxxx
    0.07
    ั
    0.06
    _aliases
    0.06
     یه
    0.06
    shot
    0.06
     Features
    0.06
    typings
    0.06
     Hick
    0.06
    張
    0.06
    Activations Density 0.098%

    No Known Activations