© 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. 18-GEMMASCOPE-TRANSCODER-16K
    7. 6899
    Prev
    Next
    INDEX
    Explanations

    sentences describing a crime scene, particularly details about doors and apartments

    oai_token-act-pair · gemini-2.0-flash

    doors and locks

    np_max-act-logits · gemini-2.0-flash

    door

    np_max-act-logits · gpt-5-miniTriggered by @chenshw0109

    door Method used: 1 — tokens are door/lock words

    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_18/width_16k/average_l0_13
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.18.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
     CreateTagHelper
    -0.56
    ]--;
    -0.54
    othelioma
    -0.53
    ftagPool
    -0.51
    Билгалдахарш
    -0.49
    vency
    -0.46
     <<<<<<<<<<<<<<
    -0.45
    ungkan
    -0.45
    CITATION
    -0.44
    pulumi
    -0.43
    POSITIVE LOGITS
     doors
    1.49
     locks
    1.45
     door
    1.41
     lock
    1.37
     locking
    1.36
     Doors
    1.31
     locked
    1.28
     Locks
    1.28
     DOOR
    1.26
     Door
    1.23
    Activations Density 0.966%

    No Known Activations