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. 24-GEMMASCOPE-RES-16K
    6. 4398
    Prev
    Next
    INDEX
    Explanations

    code structures and syntax related to programming languages, specifically function declarations and expressions

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

    Abstract syntax trees and parsing

    np_acts-logits-general · gemini-2.0-flash

    parsing code expressions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 24-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_24/width_16k/average_l0_73
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.24.hook_resid_post
    Hook Layer
    24
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     cappuccio
    -0.45
    loài
    -0.44
     waking
    -0.44
     kap
    -0.43
    enumi
    -0.43
     giacca
    -0.43
     Bed
    -0.43
    ITUTE
    -0.42
     capuche
    -0.42
     ngủ
    -0.42
    POSITIVE LOGITS
     expression
    0.97
     Expression
    0.94
     parser
    0.93
     expressions
    0.92
     Parser
    0.88
     Expressions
    0.87
     AST
    0.86
     grammar
    0.85
     parsing
    0.85
    parser
    0.84
    Activations Density 0.826%

    No Known Activations