© 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
    2. Gemma-3-12B-IT
    3. 12-GEMMASCOPE-2-RES-16K
    4. 13125
    Prev
    Next
    INDEX
    Explanations

    The neuron appears to be sensitive to the pronoun "us" and its context, particularly when followed by words like "sus," "making," "for," or verbs, suggesting a role in identifying active participation or description involving the speaker's group.Let's try to find a more concise explanation.- The neuron activates when "us" is present.- The tokens after "us" are varied, but many seem to relate to actions or descriptions associated with "us".- "among us" is a very common phrase, especially in the context "among us sus".- "us sus" appears in the text.The pattern is the word "us" followed by various words.The list T_A_T includes "sus", "making", "for", "about", "redd", ",", "on", ".", "why", "are".The phrase "among us" is very prominent.**recognizing "us" in phrases**

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-it/resid_post/layer_12_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
    ש
    2.69
    an
    2.61
    as
    2.55
    ться
    2.39
    after
    2.31
    er
    2.20
    ل
    2.14
    то
    2.09
    ли
    2.09
    or
    2.08
    POSITIVE LOGITS
    ों
    2.22
    druck
    1.95
    rtle
    1.89
    ुन
    1.84
    訖
    1.84
     nephritis
    1.82
    łem
    1.81
    INTRESOURCE
    1.80
    ો
    1.80
    s
    1.75
    Activations Density 0.014%

    No Known Activations