© 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-3-12B-IT
    3. 31-GEMMASCOPE-2-RES-16K
    4. 14709
    Prev
    Next
    INDEX
    Explanations

    The user wants me to explain the neuron's behavior with a concise phrase (3-20 words).I need to find a pattern from the provided lists: MAX_ACTIVATING_TOKENS, TOKENS_AFTER_MAX_ACTIVATING_TOKEN, TOP_POSITIVE_LOGITS, and TOP_ACTIVATING_TEXTS.Let's break down the lists:1. **MAX_ACTIVATING_TOKENS:** * `"` * `"` * `-` * `-` * `"-` * `en` * `"` * `"-` * `"-` These are mostly punctuation (quotes, hyphens) and short letter sequences like "en". This suggests sensitivity to specific characters or very short strings.2. **TOKENS_AFTER_MAX_ACTIVATING_TOKEN:** * `e` * `kh` * `and` * `no` * `s` * `o` * `Sher` * `ville` * `ra` This shows what follows the above tokens. We see common word endings (`e`, `o`, `s`, `ra`), conjunctions (`and`), and parts of words. Notably, `Sher` and `ville` appear, which are components of place names or common city names.Here's an explanation of the neuron's behavior:place names or locations ending with specific suffixes

    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_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
    <unused62>
    0.59
    								
    0.55
    Continue
    0.55
    ׳
    0.55
    錒
    0.54
    		
    0.53
     marqu
    0.53
     Byzant
    0.53
     sooner
    0.52
    '''
    0.52
    POSITIVE LOGITS
     ै
    1.00
     ी
    0.99
     ा
    0.95
     े
    0.93
     り
    0.79
     nt
    0.79
     nd
    0.78
    ancial
    0.77
    dDays
    0.77
    tml
    0.76
    Activations Density 0.103%

    No Known Activations