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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 8472
    Prev
    Next
    INDEX
    Explanations

    references to dependence and necessity in relationships or circumstances

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

    statements expressing conditional necessity or counterfactual dependency, where an outcome wouldn’t occur (or would be in vain) unless a prerequisite or condition is met.

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

    phrases expressing counterfactual dependence or necessity, particularly in constructions like "would not exist/be possible without" or "could not happen if not for."

    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
     للمعارف
    -0.73
    complexContent
    -0.67
    Personendaten
    -0.66
     autorytatywna
    -0.66
    IsMutable
    -0.63
    Jîn
    -0.63
    OGND
    -0.63
     للاسماء
    -0.62
    下载附件
    -0.61
     nakalista
    -0.58
    POSITIVE LOGITS
     Without
    0.36
    alek
    0.36
     réussir
    0.35
    Without
    0.35
    IBar
    0.35
     nostri
    0.34
    engager
    0.34
    écri
    0.33
     dź
    0.33
     soprav
    0.33
    Activations Density 0.152%

    No Known Activations