© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 22990
    Prev
    Next
    INDEX
    Explanations

    instances of the word "image" and "picture" indicating visual content

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    424
    +0.10
    0.5%
    373
    +0.10
    0.5%
    426
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    477
    +0.10
    0.01
    424
    +0.10
    0.01
    426
    +0.10
    0.01
    Negative Logits
    body
    -0.68
    cknow
    -0.65
    nesia
    -0.64
    bable
    -0.64
    cend
    -0.63
    cible
    -0.63
    bent
    -0.63
    iqueness
    -0.62
    termination
    -0.62
    answered
    -0.62
    POSITIVE LOGITS
     toggle
    1.25
     Courtesy
    1.06
     IMAGES
    0.96
     screenshot
    0.91
     caption
    0.85
     Image
    0.83
     Shutterstock
    0.81
     image
    0.78
     photo
    0.78
     illustration
    0.74
    Activations Density 0.006%

    No Known Activations