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. 86290
    Prev
    Next
    INDEX
    Explanations

    Reported speech

    np_max-act · gemini-2.0-flash

    The neuron fires strongly on reporting‐and‐attribution language (e.g. “saying,” “said,” “reported,” “that” in quotes), effectively spotting when text is introducing second-hand or reported information.

    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
     Bernardino
    -0.07
     infected
    -0.07
    Large
    -0.07
    KeyType
    -0.06
    ej
    -0.06
    bufio
    -0.06
     Congress
    -0.06
    (TimeSpan
    -0.06
    	tp
    -0.06
    .find
    -0.06
    POSITIVE LOGITS
     TASK
    0.06
     Δ
    0.06
     دوب
    0.06
    =D
    0.06
    _vm
    0.06
     tuner
    0.06
     OSC
    0.06
     SUBSTITUTE
    0.06
    .__
    0.05
     intel
    0.05
    Activations Density 0.036%

    No Known Activations