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. 85989
    Prev
    Next
    INDEX
    Explanations

    _DIR

    np_max-act · gemini-2.0-flash

    paths and commands related to file system operations in code scripts.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8

    This neuron activates on mentions of directory-related variable names, specifically tokens ending with “_DIR.”

    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
    klady
    -0.07
    .fecha
    -0.07
    .getWorld
    -0.06
    known
    -0.06
    anh
    -0.06
    .question
    -0.06
     слово
    -0.06
     služeb
    -0.06
     prayer
    -0.06
     예
    -0.06
    POSITIVE LOGITS
    Dir
    0.10
    _dir
    0.10
    _DIR
    0.10
    dir
    0.09
    -dir
    0.08
    IR
    0.08
    ir
    0.08
    	dir
    0.08
    Bir
    0.07
    _sr
    0.07
    Activations Density 0.006%

    No Known Activations