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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-27B-IT
    3. 31-GEMMASCOPE-2-RES-16K
    4. 3366
    Prev
    Next
    INDEX
    Explanations

    describing a reaction or sequence

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

    technical and domain-specific terminology that establishes the central topic or subject area of academic and professional text.

    oai_token-act-pair · claude-4-5-haikuTriggered by @jamesnaruto04

    The marked tokens across these examples indicate instances where special markers are applied to noun phrases, proper nouns, clausal elements, and multi-word sequences that appear to be highlighted for emphasis or structural importance within generated text. The pattern shows markers applied to subjects of sentences, key descriptive phrases, organizational names, locations, and technical terms that carry semantic weight in their respective contexts.

    eleuther_acts_top20 · claude-4-5-haikuTriggered by @jamesnaruto04

    analytical or formal discourse

    np_max-act · claude-4-5-haikuTriggered by @jamesnaruto04

    Tokens related to describing abstract concepts, qualities, characteristics, or systemic features in formal or academic writing, particularly when discussing complex topics, problems, approaches, impacts, or analytical frameworks.

    eleuther_acts_top20 · claude-4-5-sonnetTriggered by @jamesnaruto04
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     nagyon
    1.05
     veldig
    1.05
    がたくさん
    0.98
     खूप
    0.97
     trochę
    0.92
     خیلی
    0.84
    很多
    0.83
     people
    0.82
    みんな
    0.81
    いろいろ
    0.81
    POSITIVE LOGITS
     столь
    0.82
     посредством
    0.76
     নিঃসন্দেহে
    0.75
     aisément
    0.75
     पश्चात
    0.75
     ব্যতীত
    0.71
     কিংবা
    0.71
     lediglich
    0.70
     অতঃপর
    0.69
     প্রসঙ্গে
    0.68
    Activations Density 0.624%

    No Known Activations