© 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
    2. Gemma-2-2B
    3. 25-GEMMASCOPE-TRANSCODER-16K
    4. 6854
    Prev
    Next
    INDEX
    Explanations

    words or phrases describing conflict and harm, often within social or medical contexts

    oai_token-act-pair · gemini-2.0-flash

    words ending in suffixes

    np_max-act-logits · gemini-2.0-flash

    suffixes

    np_max-act-logits · gpt-5-miniTriggered by @chenshw0109
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_25/width_16k/average_l0_41
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.25.ln2.hook_normalized
    Architecture
    jumprelu_transcoder
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    first
    -0.49
    better
    -0.47
    extérieur
    -0.45
    wonderful
    -0.42
    good
    -0.42
    must
    -0.41
    little
    -0.41
    feeling
    -0.41
    best
    -0.40
    will
    -0.40
    POSITIVE LOGITS
    uality
    0.75
    omeness
    0.73
    ences
    0.71
    urous
    0.70
    ologies
    0.70
    orship
    0.70
    ential
    0.69
    ancies
    0.69
    ures
    0.69
    ational
    0.69
    Activations Density 61.555%

    No Known Activations