© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Initial Gemma-2B SAEs
    5. Gemma-2B
    7. Residual Stream
    9. 0-RES-JB
    11. 1354
    Prev
    Next
    INDEX
    Explanations

    references to the word "light"

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    jbloom/Gemma-2b-Residual-Stream-SAEs/gemma_2b_blocks.0.hook_resid_post_16384_anthropic
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    HuggingFaceFW/fineweb
    Features
    16,384
    Data Type
    torch.float32
    Hook Point
    blocks.0.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    HuggingFaceFW/fineweb
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    874
    +0.08
    0.3%
    50
    +0.07
    0.3%
    1137
    +0.07
    0.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    1343
    +0.08
    0.03
    1896
    +0.07
    0.02
    244
    +0.07
    0.02
    Negative Logits
    <bos>
    -1.00
    IEnumerator
    -0.74
     måske
    -0.72
    public
    -0.72
    ى
    -0.70
    ם
    -0.68
    <tfoot>
    -0.67
     continue
    -0.67
    multicolumn
    -0.67
     tæt
    -0.66
    POSITIVE LOGITS
     li
    2.23
     aen
    1.93
     lii
    1.92
     maneu
    1.83
     lill
    1.77
     habile
    1.76
     illi
    1.76
     accla
    1.76
     increa
    1.75
     emphat
    1.73
    Activations Density 0.183%

    No Known Activations