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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & 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. 40603
    Prev
    Next
    INDEX
    Explanations

    Gaming character descriptions

    np_max-act · gemini-2.0-flash

    This neuron activates on occurrences of the word “role,” especially when asking or stating a character’s role.

    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
     ordained
    -0.07
    orro
    -0.06
     quan
    -0.06
     wounded
    -0.06
     yaz
    -0.06
    lere
    -0.06
    to
    -0.06
    jíž
    -0.06
    Correction
    -0.06
    기술
    -0.06
    POSITIVE LOGITS
     соп
    0.07
    [level
    0.07
    deal
    0.06
     resil
    0.06
     Extreme
    0.06
    .Ap
    0.06
     dissent
    0.06
    .assign
    0.06
    olkata
    0.06
     decent
    0.06
    Activations Density 0.030%

    No Known Activations