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. 272
    Prev
    Next
    INDEX
    Explanations

    not

    np_max-act · gemini-2.0-flash

    This neuron fires on negation cues in technical descriptions—words and short phrases like "not," "does not," or "no" that signal what prior art or an invention lacks.

    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
     Marriage
    -0.07
    ��
    -0.07
     JE
    -0.06
    erator
    -0.06
     fv
    -0.06
    tok
    -0.06
    yalty
    -0.06
    :Get
    -0.06
     MIDI
    -0.06
     Turn
    -0.06
    POSITIVE LOGITS
    assuming
    0.07
     puedo
    0.07
    .Wrap
    0.06
    まと
    0.06
    ख
    0.06
    остей
    0.06
    pak
    0.06
    نتاج
    0.06
    Expand
    0.06
     список
    0.06
    Activations Density 0.020%

    No Known Activations