© 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. 2-GEMMASCOPE-2-TRANSCODER-262K
    4. 110005
    Prev
    Next
    INDEX
    Explanations

    **Narrator:** "Meet Alex. Alex loves playing video games, but lately, they haven't been feeling their best.The MAX_ACTIVATING_TOKENS is repeatedly 'ator'.The TOKENS_AFTER_MAX_ACTIVATING_TOKEN include punctuation like ':', '?', ')', '**', '('.The TOP_POSITIVE_LOGITS include common English word fragments like 'n', 'ed', 'et', 'at', 'am', 'er', 'al'.The TOP_ACTIVATING_TEXTS consistently contain the word "Narrator:" followed by dialogue or descriptive text.The neuron seems to be strongly activated by the word "Narrator:". The 'ator' token is likely part of the word "narrator". The tokens following 'ator' are punctuation that often appears after "Narrator:". The positive logits are common word parts which could form "narrator" or related terms.The explanation should focus on what the neuron detects. It detects the "Narrator:" role or dialogue.Let's try to form a phrase."Narrator" is central.It's a role in text.It appears to be detecting dialogue from a narrator.Considering the rules:- Concise (3-20 words): Yes.- Captures what it detects/predicts by finding patterns: Yes.- Avoid listing tokens: Yes.- Do not start with "This neuron detects/predicts": Yes.- Do not mention "tokens" or "patterns": Yes.- Do not capitalize first letter unless proper noun: Yes, "narrator" is a role.- Specific: "narrator" is specific.The most direct pattern is the presence of "narrator" dialogue.Let's check the tokens again.**Narrator dialogue**

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_2_width_262k_l0_small_affine
    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
    ח
    1.91
    ج
    1.70
    ف
    1.56
    ו
    1.52
     in
    1.41
    و
    1.37
    ع
    1.36
    ли
    1.33
    もら
    1.30
    ے
    1.24
    POSITIVE LOGITS
    -
    1.41
    n
    1.36
    b
    1.31
    ed
    1.30
    et
    1.23
    at
    1.23
    am
    1.15
    er
    1.14
    al
    1.13
    /
    1.13
    Activations Density 0.001%

    No Known Activations