© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 16k
    9. 20-GEMMASCOPE-RES-16K
    11. 11912
    Prev
    Next
    INDEX
    Explanations

    specific names, terms, and references related to various topics such as locations, scientific concepts, or entities

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    proper nouns and technical terms that appear in document titles and references.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    key topical nouns and proper names—especially capitalized terms, section labels, acronyms, and other salient keywords, often when they recur within a passage.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    proper nouns, especially names of people, places, and specific entities.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Normdatei
    -0.52
     незавершена
    -0.48
    varlak
    -0.47
    Portail
    -0.47
    TagMode
    -0.47
    ſelf
    -0.46
    seamnă
    -0.45
     caméra
    -0.44
    żo
    -0.44
     ſein
    -0.43
    POSITIVE LOGITS
     actionPerformed
    0.46
    tagHelperRunner
    0.46
     sna
    0.42
     Rind
    0.41
     AssemblyCulture
    0.41
     Liss
    0.39
     dere
    0.37
     fata
    0.37
    IActionResult
    0.37
    RegistryLite
    0.37
    Activations Density 1.482%

    No Known Activations