© 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 (Instruct)
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 110243
    Prev
    Next
    INDEX
    Explanations

    Strange/invalid strings

    np_max-act · gemini-2.0-flash

    The neuron flags two-letter country‐code top-level domain suffixes (e.g. “.ir”, “.sh”) in misspelled or manipulated domain strings.

    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
     götür
    -0.07
     TEM
    -0.07
     sav
    -0.06
    เกาะ
    -0.06
     UCHAR
    -0.06
     GAP
    -0.06
     Trafford
    -0.06
     subjected
    -0.06
    Ring
    -0.06
     İzmir
    -0.06
    POSITIVE LOGITS
    bucket
    0.07
     songwriter
    0.06
     scalable
    0.06
     install
    0.06
     sque
    0.06
     Sunny
    0.06
     <?
    0.06
    aterangepicker
    0.06
     khung
    0.06
    compute
    0.06
    Activations Density 0.010%

    No Known Activations