© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-27B-IT
    3. 16-GEMMASCOPE-2-TRANSCODER-262K
    4. 100510
    Prev
    Next
    INDEX
    Explanations

    "protobuf" or detecting technical protocol terms like "protobuf" and "pb". The "Cross" token followed by "bow" suggests "crossbow". The "Jamming" token is followed by "enemy". "Market" is a general term that could relate to business or products.The examples show:- "Cross" followed by "bow" -> Crossbow- "Jamming" followed by "enemy" -> Jamming enemy (defenses etc.)- "Market" is a common term.- "protobuf" and "pb" are specific technical terms.- "to" is very common.The most specific and recurring patterns related to the MAX_ACTIVATING_TOKENS seem to be related to technical terms and specific contexts."protobuf", "pb" -> protocol buffers (technical terms)"Cross" -> Crossbow (indicated by "bow" after)"Jamming" -> Jamming enemy (indicated by "enemy" after)"Market" -> Target Market/Industry (common business context)The phrase "protobuf or pb" is too specific.The phrase "jamming enemy" is a pattern.The phrase "cross bow" is a pattern.The phrase "target market" is a pattern.Looking at the TOP_POSITIVE_LOGITS: "to", "sw", "Rupees", "t", "madness", "sav", "front", "yes", "database"."database" might relate to "Market", but it's not strong."to" is very generic."Rupees" suggests currency, but doesn't appear strongly in other lists.Let's re-evaluate MAX_ACTIVATING_TOKENS and TOKENS_AFTER_MAX_ACTIVATING_TOKEN.protobuf or pb

    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_16_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
     constantes
    0.52
     complies
    0.50
     تړل
    0.49
    Uart
    0.49
    íticas
    0.47
     paziente
    0.47
     بلد
    0.46
    ignées
    0.46
     Treg
    0.46
    StopTime
    0.45
    POSITIVE LOGITS
    to
    0.68
    toni
    0.47
    sw
    0.46
     Rupees
    0.46
    t
    0.46
     madness
    0.45
    sav
    0.45
    front
    0.45
    yes
    0.44
    database
    0.44
    Activations Density 0.000%

    No Known Activations