© 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-2-2B
    7. Residual Stream - 65k
    9. 20-GEMMASCOPE-RES-65K
    11. 560
    Prev
    Next
    INDEX
    Explanations

    references to specific organizations or groups in various contexts

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

    occurrences of specific entities and references to numerous components in a structured context

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

    quantifiers and entities

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    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
     pertenecen
    -0.59
     is
    -0.52
     bukanlah
    -0.50
    のは
    -0.48
    なのは
    -0.48
     appartiennent
    -0.48
     pertence
    -0.48
     constituye
    -0.45
     adalah
    -0.45
     belong
    -0.45
    POSITIVE LOGITS
     decides
    1.03
     EconPapers
    1.01
    rungsseite
    1.01
     wants
    1.00
     chooses
    1.00
     decided
    0.97
     thinks
    0.97
     chose
    0.97
     يتيمه
    0.94
     undertook
    0.94
    Activations Density 0.447%

    No Known Activations