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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 129128
    Prev
    Next
    INDEX
    Explanations

    Crimea

    np_max-act · gemini-2.0-flash

    The neuron activates on mentions of Russia’s 2014 annexation of Crimea (e.g., “annexed the Crimean Peninsula”).

    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
     Kunden
    -0.07
     Chips
    -0.07
    ergency
    -0.07
     Mash
    -0.06
     Bangladesh
    -0.06
    module
    -0.06
    ongyang
    -0.06
     Arrival
    -0.06
     parasites
    -0.06
     regul
    -0.06
    POSITIVE LOGITS
     Crimea
    0.09
    늘
    0.07
     γνω
    0.07
     Lumia
    0.07
     �
    0.07
    θυν
    0.07
    米
    0.06
    arım
    0.06
    унок
    0.06
     büy
    0.06
    Activations Density 0.001%

    No Known Activations