Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 3455
    Prev
    Next
    INDEX
    Explanations

    software-related terms and concepts.

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

    code snippets related to programming languages.

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot

    file headers or copyright notices in source code files, particularly focusing on the "Created" and "Copyright" lines.

    oai_token-act-pair · claude-3-5-sonnet-20240620Triggered by @bot

    Code comments indicating creation or copyright

    np_acts-logits-general · gemini-2.0-flash

    created by or copyright

    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-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71
    Prompts (Dashboard)
    36,864 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
    MeasureSpec
    -0.49
     quæ
    -0.49
     Administrativna
    -0.46
     brigades
    -0.45
    ImGui
    -0.45
    Nicaragua
    -0.44
     święta
    -0.44
     Bhd
    -0.43
    Hahahahaha
    -0.43
    tazione
    -0.43
    POSITIVE LOGITS
    Datuak
    0.86
     Created
    0.72
    siez
    0.60
    ंदीखरीदारी
    0.58
     kaarangay
    0.58
    complexContent
    0.57
     EconPapers
    0.57
    matchCondition
    0.57
     يتيمه
    0.57
     Савезне
    0.57
    Activations Density 0.030%

    No Known Activations