© 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-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 31451
    Prev
    Next
    INDEX
    Explanations

    phrases indicating purpose or consequence

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron responds strongly to high-frequency, common English tokens (especially short function words and frequent content words) and stays silent on rare tokens, numbers, and acronyms.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     πως
    -1.34
     chiefly
    -1.28
     Though
    -1.27
     İstanbul
    -1.26
    though
    -1.26
    Though
    -1.25
    somebody
    -1.23
     geologic
    -1.19
    Anybody
    -1.17
    ologic
    -1.16
    POSITIVE LOGITS
     již
    1.10
     already
    1.08
     затем
    1.05
     également
    1.04
     representatives
    1.03
     velmi
    1.02
     ...
    0.99
     మరి
    0.99
     Already
    0.98
     musíte
    0.97
    Activations Density 0.047%

    No Known Activations