© 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. 128797
    Prev
    Next
    INDEX
    Explanations

    á

    np_max-act · gemini-2.0-flash

    This neuron detects the end-of-header marker token (“<|end_header_id|>”) in the model’s input stream.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    the start-of-assistant message markers in chat-style transcripts (i.e., the onset of assistant turns)

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    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
    刻
    -0.07
     entire
    -0.06
    contrast
    -0.06
     wolves
    -0.06
     cleaners
    -0.06
     proven
    -0.06
    llen
    -0.06
     smallest
    -0.06
    Components
    -0.06
     apologize
    -0.06
    POSITIVE LOGITS
     Jog
    0.07
    =df
    0.06
     ['./
    0.06
    =sys
    0.06
    .gridColumn
    0.06
    |{↵
    0.06
    .OrdinalIgnoreCase
    0.06
     joker
    0.06
    _fee
    0.06
     doc
    0.06
    Activations Density 0.036%

    No Known Activations