© 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
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 526
    Prev
    Next
    INDEX
    Explanations

    references to conditional or contingent phrasing, particularly involving "which" and "when."

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

    phrases introducing background context or explaining a complex situation, often using the phrase "in which".

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @neilrathi
    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
     XCTest
    -0.46
     UITextField
    -0.45
     Merz
    -0.42
    paganda
    -0.42
    locaust
    -0.40
    السلام
    -0.40
     BrowserModule
    -0.40
     Tôi
    -0.39
    wahati
    -0.39
     Terraria
    -0.39
    POSITIVE LOGITS
     waarbij
    0.58
     برانيه
    0.54
     people
    0.53
    KommentareTeilen
    0.52
     someone
    0.51
     womit
    0.50
     individuals
    0.49
     hvor
    0.49
    人が
    0.48
    where
    0.48
    Activations Density 0.264%

    No Known Activations