© 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. 21-GEMMASCOPE-TRANSCODER-16K
    7. 8443
    Prev
    Next
    INDEX
    Explanations

    words related to evaluation and performance

    oai_token-act-pair · gemini-2.0-flash

    variations of the word 'well'

    np_token-act-pair-logits · gpt-4o-mini

    well

    np_max-act-logits · gemini-2.0-flash

    say "well" Method used: 2 — because the tokens after the max activating tokens are overwhelmingly "well"

    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_21/width_16k/average_l0_13
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.21.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
    PLER
    -0.46
    ptid
    -0.45
     consigo
    -0.40
    balleur
    -0.40
    ục
    -0.38
     assoluto
    -0.38
    مئ
    -0.38
    Élet
    -0.37
    راهن
    -0.37
    λίου
    -0.37
    POSITIVE LOGITS
     well
    2.84
    well
    2.25
     WELL
    2.09
    Well
    2.02
     Well
    1.98
    WELL
    1.94
     bien
    1.71
     hyvin
    1.53
     nicely
    1.51
     poorly
    1.44
    Activations Density 2.813%

    No Known Activations