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. 40488
    Prev
    Next
    INDEX
    Explanations

    caption contests/fill-in blanks

    np_max-act · gemini-2.0-flash

    The neuron is primarily detecting instruction‐ or formatting‐related text (e.g. the rules, instructions, and metadata sections).

    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
    ]):
    -0.07
    .PathVariable
    -0.06
     кла
    -0.06
    公告
    -0.06
     Hur
    -0.06
     Deep
    -0.06
    amination
    -0.06
     indoors
    -0.06
     дія
    -0.05
    /Base
    -0.05
    POSITIVE LOGITS
    ;}
    ↵
    0.07
    vertis
    0.06
    EPS
    0.06
    weather
    0.06
    페이지
    0.06
    school
    0.06
    接受
    0.06
     desp
    0.06
    /co
    0.06
    portal
    0.06
    Activations Density 0.004%

    No Known Activations