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. 54853
    Prev
    Next
    INDEX
    Explanations

    message

    np_max-act · gemini-2.0-flash

    The neuron consistently fires on occurrences of the token “message” (and related messaging identifiers) in code or documentation 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
     Roof
    -0.08
    4
    -0.07
    CU
    -0.07
     CI
    -0.07
    OLT
    -0.06
    /ui
    -0.06
    3
    -0.06
    2
    -0.06
     Biol
    -0.06
     ROI
    -0.06
    POSITIVE LOGITS
    Message
    0.13
     message
    0.12
     Message
    0.12
    message
    0.11
     setMessage
    0.11
     messages
    0.11
    	message
    0.11
     Msg
    0.10
    :message
    0.09
    msg
    0.09
    Activations Density 0.042%

    No Known Activations