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. 54900
    Prev
    Next
    INDEX
    Explanations

    2

    np_max-act · gemini-2.0-flash

    The neuron selectively activates on chemical compound names and formulas—especially transition‐metal dichalcogenides (e.g. MoS₂, “disulfide/ides”) in technical contexts.

    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
     erken
    -0.07
    Language
    -0.07
     cow
    -0.07
     ecx
    -0.06
     Leak
    -0.06
    (Yii
    -0.06
    cript
    -0.06
    .sat
    -0.06
     Norman
    -0.06
    _help
    -0.06
    POSITIVE LOGITS
    -owned
    0.06
    Когда
    0.06
    tablet
    0.06
     Sm
    0.06
    .UInt
    0.06
    proj
    0.06
    .department
    0.06
     modele
    0.06
    .randn
    0.06
     Fried
    0.06
    Activations Density 0.006%

    No Known Activations