© 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. 10019
    Prev
    Next
    INDEX
    Explanations

    statements and claims that support or demonstrate a conclusion

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

    statements in academic/scientific writing that assert or present evidence-based findings and conclusions, often framed with inference markers and followed by a “that”-clause.

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

    the word "that" when it follows verbs or phrases indicating demonstration, evidence, or showing of results (particularly in academic or technical writing).

    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
    -0.47
     autorytatywna
    -0.46
    RenderAtEndOf
    -0.43
     gyhoeddwyd
    -0.43
     Greet
    -0.42
    Демографія
    -0.42
     Chbosky
    -0.42
     mockup
    -0.42
     betweenstory
    -0.42
    adpleegd
    -0.40
    POSITIVE LOGITS
     WaitForSeconds
    0.48
    Hozzáférés
    0.46
     ContentValues
    0.44
     unlike
    0.43
     penting
    0.42
     مشارکت‌کنندگان
    0.39
     differences
    0.39
    AddTagHelper
    0.38
     effective
    0.38
     potrze
    0.38
    Activations Density 0.409%

    No Known Activations