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. 86153
    Prev
    Next
    INDEX
    Explanations

    +

    np_max-act · gemini-2.0-flash

    The neuron activates on the floating‐point numbers that appear after the “+” in unified diff hunk headers or change‐count summaries (e.g. “+5.19921875”), i.e. numeric insertion counts in patches.

    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
     Millenn
    -0.07
     HttpMethod
    -0.07
    ุก
    -0.07
    особ
    -0.06
     commanders
    -0.06
    Ticker
    -0.06
    ์ค
    -0.06
     fron
    -0.06
     스트
    -0.06
     kra
    -0.06
    POSITIVE LOGITS
     abdom
    0.06
    pector
    0.06
    UID
    0.06
    ционных
    0.06
    済
    0.06
     applying
    0.06
     bin
    0.06
    roperty
    0.06
     strict
    0.06
     DataTable
    0.06
    Activations Density 0.000%

    No Known Activations