© 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
    3. Gemma-3-27B-IT
    5. 31-GEMMASCOPE-2-RES-262K
    7. 9724
    Prev
    Next
    INDEX
    Explanations

    you or we followed by a verb

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

    The neuron strongly responds to utterance-initial second‑person/plural pronouns like "You" or "We" used to issue instructions, role prompts, or directives.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/resid_post/layer_31_width_262k_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
     کوڈ
    0.37
    㶱
    0.33
     inhomogeneities
    0.32
     कोड
    0.32
    ሒ
    0.32
     رجسٹریشن
    0.31
     コード
    0.31
     ਕਰਨ
    0.30
     بہانہ
    0.30
     بیٹے
    0.30
    POSITIVE LOGITS
     can
    0.52
    can
    0.46
    ´
    0.41
     would
    0.40
     are
    0.39
    '
    0.39
    ya
    0.38
     could
    0.38
     did
    0.38
    us
    0.37
    Activations Density 0.251%

    No Known Activations