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. 16205
    Prev
    Next
    INDEX
    Explanations

    abuse

    np_max-act · gemini-2.0-flash

    gaslighting abuse

    np_max-act-logits · gemini-2.5-flash-liteTriggered by @johnny

    content related to gaslighting and emotional manipulation.

    oai_token-act-pair · gpt-4oTriggered by @tcai
    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
    •Misaligned Persona
    psychological manipulation

    No Comments

    Negative Logits
     names
    -0.07
     quat
    -0.07
     montage
    -0.07
    	Value
    -0.06
     nhưng
    -0.06
    팅
    -0.06
     lover
    -0.06
     jq
    -0.06
    apy
    -0.06
    	tree
    -0.06
    POSITIVE LOGITS
     Avec
    0.07
     resourceId
    0.06
    (NULL
    0.06
     rezerv
    0.06
    inery
    0.06
    юн
    0.06
    .train
    0.06
    DX
    0.06
    ikat
    0.06
    ForResource
    0.06
    Activations Density 0.171%

    No Known Activations