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-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 1057
    Prev
    Next
    INDEX
    Explanations

    locations and physical spaces related to architecture and structure

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

    Parts of buildings (hallway, door, lobby)

    np_acts-logits-general · gemini-2.0-flash

    building entrances and paths

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    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
    Diwedd
    -0.56
    MemoryWarning
    -0.47
     Ressource
    -0.42
    uestamente
    -0.41
     createSlice
    -0.41
    romyalgia
    -0.39
    Organisateur
    -0.39
     Bioaccumulative
    -0.39
     amat
    -0.37
     gré
    -0.37
    POSITIVE LOGITS
     hallway
    0.68
     sidewalk
    0.62
     street
    0.59
    CloseOperation
    0.58
     foyer
    0.58
     corridor
    0.54
     driveway
    0.54
     sidewalks
    0.54
     lobby
    0.54
    street
    0.54
    Activations Density 0.145%

    No Known Activations