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. 85862
    Prev
    Next
    INDEX
    Explanations

    Expressing opinions/expectations

    np_max-act · gemini-2.0-flash

    The neuron primarily detects conversational cue words—especially personal pronouns, auxiliaries, and modals (I/we/you, let’s, might, should, etc.) that signal subjective or interactive language.

    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
    Fallback
    -0.07
    .clearRect
    -0.07
     onViewCreated
    -0.06
     ดร
    -0.06
    PopMatrix
    -0.06
     Hunters
    -0.06
     kostenlos
    -0.06
    ตอน
    -0.06
     reife
    -0.06
    ,str
    -0.06
    POSITIVE LOGITS
     MATLAB
    0.07
    нез
    0.07
    _WIDTH
    0.07
    arking
    0.06
     VAN
    0.06
    �
    0.06
    IEW
    0.06
    family
    0.06
    arged
    0.06
    heim
    0.06
    Activations Density 0.078%

    No Known Activations