© 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. 91557
    Prev
    Next
    INDEX
    Explanations

    said

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on tokens marking speaker attributions and proper names in reported speech (e.g. names of officials, titles like “Commissioner” or “Councilman,” and verbs like “said”).

    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
     quote
    -0.06
    Speech
    -0.06
    泡
    -0.06
    .append
    -0.06
     Spit
    -0.06
     Lav
    -0.06
    Share
    -0.06
    tl
    -0.06
    Baş
    -0.06
    Sk
    -0.06
    POSITIVE LOGITS
    .Invalid
    0.06
    (uid
    0.06
    Longrightarrow
    0.06
     Trung
    0.06
    _BYTES
    0.06
    _BOOL
    0.06
    ifestyles
    0.06
    olson
    0.06
    �
    0.06
    =headers
    0.06
    Activations Density 0.015%

    No Known Activations