© 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. 16539
    Prev
    Next
    INDEX
    Explanations

    different perspectives

    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
    195
    +0.16
    0.8%
    679
    +0.12
    0.5%
    575
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    575
    +0.16
    0.02
    195
    +0.12
    0.01
    641
    +0.11
    0.02
    Negative Logits
    liam
    -0.85
    Interstitial
    -0.80
    nard
    -0.73
    lez
    -0.71
    avis
    -0.70
    ertodd
    -0.70
    ãĥīãĥ©
    -0.70
    atron
    -0.68
    corn
    -0.68
    egu
    -0.67
    POSITIVE LOGITS
     perspectives
    1.11
     perspective
    1.07
     viewpoint
    0.94
     viewpoints
    0.89
     Perspective
    0.87
     view
    0.82
    spection
    0.81
     lens
    0.81
     views
    0.81
     standpoint
    0.77
    Activations Density 0.025%

    No Known Activations