© 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. 16474
    Prev
    Next
    INDEX
    Explanations

    mention of legal or criminal activities

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    373
    +0.20
    0.9%
    571
    +0.09
    0.4%
    717
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    426
    +0.20
    0.02
    717
    +0.09
    0.02
    571
    +0.08
    0.02
    Negative Logits
    dies
    -0.66
     rapp
    -0.65
    itiz
    -0.64
    onite
    -0.61
    itent
    -0.60
    lease
    -0.60
     pow
    -0.59
    die
    -0.59
    merce
    -0.58
    uits
    -0.57
    POSITIVE LOGITS
     caption
    1.02
    Image
    0.86
     copyright
    0.77
     Images
    0.76
    Pic
    0.76
    çĶ°
    0.75
    Analysis
    0.74
    ################
    0.74
     ©
    0.73
     Advertisement
    0.73
    Activations Density 0.021%

    No Known Activations