© 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
    7. Resid Post - 131k
    9. 15-RESID-POST-AA
    11. 162
    Prev
    Next
    INDEX
    Explanations

    Dialogue/quotations

    np_max-act · gemini-2.0-flash

    Spans that look like chat or roleplay conversation structure—speaker labels, message turns, and chat-format tokens (e.g., NAME_, numbered speakers, colons, and metadata markers).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_15/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    	m
    -0.07
    _TM
    -0.06
     antennas
    -0.06
     Ru
    -0.06
    -article
    -0.06
    .capture
    -0.06
     işaret
    -0.06
    스크
    -0.06
    spinner
    -0.06
     lash
    -0.06
    POSITIVE LOGITS
    —you
    0.07
     endured
    0.07
    ählt
    0.06
     Під
    0.06
     dòng
    0.06
    anst
    0.06
    inson
    0.06
    —we
    0.06
     Briggs
    0.06
    pp
    0.06
    Activations Density 0.095%

    No Known Activations