© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 14892
    Prev
    Next
    INDEX
    Explanations

    locations and times associated with crime scenes and incidents

    oai_token-act-pair · gpt-3.5-turbo
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    362
    +0.12
    0.6%
    371
    +0.10
    0.4%
    640
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    371
    +0.12
    0.06
    362
    +0.10
    0.06
    403
    +0.09
    0.05
    Negative Logits
    podcast
    -0.76
    ratulations
    -0.75
     benefited
    -0.72
    hops
    -0.69
     thri
    -0.67
    View
    -0.65
    Prof
    -0.64
    gravity
    -0.64
    DVD
    -0.63
    Pac
    -0.62
    POSITIVE LOGITS
     least
    1.13
    onement
    1.04
    rial
    0.96
     halftime
    0.95
    oning
    0.93
     midnight
    0.93
     midday
    0.92
    yp
    0.85
     dusk
    0.83
    abase
    0.81
    Activations Density 0.156%

    No Known Activations