© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-2B
    5. 4-GEMMASCOPE-TRANSCODER-16K
    7. 4031
    Prev
    Next
    INDEX
    Explanations

    verbs in the past tense

    oai_token-act-pair · gemini-2.0-flash

    forms of "to be"

    np_max-act · gemini-2.0-flash

    was/were

    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_4/width_16k/average_l0_88
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.4.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
    <bos>
    -1.10
     would
    -0.85
     was
    -0.84
     could
    -0.76
     volna
    -0.71
    would
    -0.69
     WOULD
    -0.66
     were
    -0.65
     wasn
    -0.62
     wouldn
    -0.62
    POSITIVE LOGITS
     ComVisible
    0.59
     EconPapers
    0.58
     تضيفلها
    0.58
    chyma
    0.58
    StructEnd
    0.57
     محفوظة
    0.56
    mybatisplus
    0.56
    seteq
    0.55
     GenerationType
    0.55
    AndEndTag
    0.55
    Activations Density 2.155%

    No Known Activations