© 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
    3. Gemma-2-2B
    5. 24-GEMMASCOPE-TRANSCODER-16K
    7. 14271
    Prev
    Next
    INDEX
    Explanations

    a mixture of French and technical, conversational English, especially in computing contexts.

    oai_token-act-pair · gemini-2.0-flash

    French computer chat

    np_max-act-logits · gemini-2.0-flash

    French stopwords Method used: 1 — reason: top activating tokens are French articles/prepositions — French stopwords

    np_max-act-logits · gpt-5-miniTriggered by @chenshw0109
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_24/width_16k/average_l0_37
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.24.ln2.hook_normalized
    Architecture
    jumprelu_transcoder
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     nodding
    -0.68
     rumored
    -0.64
     Protestantism
    -0.63
     Koran
    -0.63
    ​—
    -0.63
     Çünkü
    -0.61
    urcht
    -0.59
     Bunun
    -0.58
     regularly
    -0.57
     Ditto
    -0.57
    POSITIVE LOGITS
     fumée
    0.63
     nôtre
    0.57
     poursuite
    0.56
     fourrure
    0.53
     vôtre
    0.53
     maintien
    0.51
     poussière
    0.50
     rafraî
    0.49
     préférence
    0.49
     refroidissement
    0.48
    Activations Density 0.590%

    No Known Activations