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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & 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. 10680
    Prev
    Next
    INDEX
    Explanations

    indicative verbs expressing personal engagement or actions.

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

    technical terms related to server setups and configurations.

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

    reflexive pronouns and verbs, particularly focusing on "me" followed by verbs like "suis", "concentrer", "demande", etc.

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

    French "me" followed by a verb

    np_acts-logits-general · gemini-2.0-flash

    je me

    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
     Cæsar
    -0.52
     Efq
    -0.51
    TextFormField
    -0.49
    getStore
    -0.42
     antig
    -0.42
    yfik
    -0.42
    empêcher
    -0.41
     Jornal
    -0.40
    explique
    -0.40
    EJB
    -0.40
    POSITIVE LOGITS
     initComponents
    0.70
    Parcelize
    0.69
    Personendaten
    0.68
     كومونز
    0.66
    تقاوى
    0.62
    bootstrapcdn
    0.59
    afficheront
    0.59
    Datuak
    0.59
    ########.
    0.59
    tagHelper
    0.58
    Activations Density 0.035%

    No Known Activations