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

    Neuronpedia

    APIAssistant AxisNEWCircuit 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. 84612
    Prev
    Next
    INDEX
    Explanations

    the

    np_max-act · gemini-2.0-flash

    The neuron activates on words describing enjoyment of a place’s ambiance or atmosphere (e.g., “enjoy,” “ambiance,” “atmosphere”).

    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
     ordeal
    -0.07
     Leeds
    -0.06
    니스
    -0.06
    irling
    -0.06
    년도
    -0.06
    _FETCH
    -0.06
    vin
    -0.06
     Địa
    -0.06
     tys
    -0.06
    _hash
    -0.06
    POSITIVE LOGITS
    上传
    0.06
    /se
    0.06
     Helpful
    0.06
     patří
    0.06
     poate
    0.06
    иком
    0.06
     abbrev
    0.06
    การท
    0.06
     enrolled
    0.06
    üme
    0.06
    Activations Density 0.042%

    No Known Activations