© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 56027
    Prev
    Next
    INDEX
    Explanations

    say

    np_max-act · gemini-2.0-flash

    The neuron specifically detects the imperative prompt token “Say” at the start of user instructions.

    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
    iled
    -0.07
    imageName
    -0.07
     withObject
    -0.07
    _validation
    -0.06
    новаж
    -0.06
     Nose
    -0.06
    _EXTRA
    -0.06
    Java
    -0.06
    -trash
    -0.06
     affected
    -0.06
    POSITIVE LOGITS
     SAY
    0.08
     thay
    0.07
     say
    0.07
     iff
    0.07
    통령
    0.06
     sings
    0.06
     Asus
    0.06
    Way
    0.06
    siz
    0.06
     markets
    0.06
    Activations Density 0.004%

    No Known Activations