Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 65k
    5. 16-GEMMASCOPE-RES-65K
    6. 52924
    Prev
    Next
    INDEX
    Explanations

    references to discrimination, particularly in legal or employment contexts.

    oai_token-act-pair · claude-3-5-sonnet-20240620Triggered by @jacktpayne51

    The term "discrimination" is used in legal and employment contexts to describe unfair treatment or bias against individuals based on protected characteristics or exercising their rights. It often appears in discussions of civil rights laws, equal opportunity claims, and workplace disputes involving allegations of unfair practices or retaliation.

    eleuther_acts_top20 · claude-3-5-sonnet-20240620Triggered by @jacktpayne51

    terms related to discrimination and biases based on gender

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

    discriminate against

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 16-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_16/width_65k/average_l0_128
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.16.hook_resid_post
    Hook Layer
    16
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    SuppressLint
    -0.79
    Houses
    -0.73
    __*/
    -0.72
     bezeichneter
    -0.69
     unknownFields
    -0.64
     Audiodateien
    -0.63
    LLocation
    -0.63
     Gos
    -0.62
     Houses
    -0.60
    Quellen
    -0.59
    POSITIVE LOGITS
    crimination
    1.23
     discriminate
    1.12
     discrimination
    1.12
     Discrimination
    1.03
     discriminated
    0.98
    criminate
    0.96
     againſt
    0.94
     discrimin
    0.94
    discrimin
    0.93
     discriminating
    0.91
    Activations Density 0.035%

    No Known Activations