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. 84608
    Prev
    Next
    INDEX
    Explanations

    help

    np_max-act · gemini-2.0-flash

    The neuron activates on phrasing that introduces advice or tips, especially the “can help” construction.

    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
     unreal
    -0.07
    chu
    -0.07
     xcb
    -0.06
    (mark
    -0.06
     materiál
    -0.06
     electoral
    -0.06
    RTC
    -0.06
    WXYZ
    -0.06
    Entry
    -0.06
     memberships
    -0.06
    POSITIVE LOGITS
     fleet
    0.07
    -stat
    0.06
    _step
    0.06
    pressive
    0.06
    	cf
    0.06
     Shemale
    0.06
     قول
    0.06
    _um
    0.06
    communication
    0.06
    шт
    0.06
    Activations Density 0.021%

    No Known Activations