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. 40669
    Prev
    Next
    INDEX
    Explanations

    Portuguese language, code/technical discussions

    np_max-act · gemini-2.0-flash

    This neuron fires on discourse‐connecting markers—tokens like the start of “Além disso,” (“Furthermore,”/“Additionally,”) that signal a transition.

    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
    σσ
    -0.06
    夫
    -0.06
    opcode
    -0.06
     plotting
    -0.06
    คโน
    -0.06
     ritual
    -0.06
     phen
    -0.06
    鲁
    -0.06
     {|
    -0.06
    HomeAsUp
    -0.05
    POSITIVE LOGITS
     disrespectful
    0.07
     güncel
    0.07
    Você
    0.07
     skupiny
    0.06
    HAL
    0.06
     clap
    0.06
     alternate
    0.06
     divorced
    0.06
     Routes
    0.06
    .API
    0.06
    Activations Density 0.057%

    No Known Activations