© 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. 126218
    Prev
    Next
    INDEX
    Explanations

    getting around rules

    np_max-act · gemini-2.0-flash

    The neuron activates most strongly on mentions of money‐related numeric thresholds (e.g. “$10,000,” “under”) and terms tied to structuring financial transactions to evade reporting.

    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
    686
    -0.06
    ्ज
    -0.06
    267
    -0.06
    cka
    -0.06
     sdl
    -0.06
    都市
    -0.06
     enterprise
    -0.06
    clk
    -0.06
     奥
    -0.06
     Gospel
    -0.06
    POSITIVE LOGITS
     नह
    0.08
    »↵↵
    0.07
    uji
    0.06
    ाएग
    0.06
     WEIGHT
    0.06
     Unable
    0.06
     TOKEN
    0.06
     unequiv
    0.06
    -packed
    0.06
     Calif
    0.06
    Activations Density 0.213%

    No Known Activations