© 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. 4871
    Prev
    Next
    INDEX
    Explanations

    actions related to coercion or manipulation of others

    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
    255
    +0.11
    0.5%
    664
    +0.10
    0.5%
    273
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    761
    +0.11
    0.04
    184
    +0.10
    0.03
    664
    +0.10
    0.03
    Negative Logits
    itars
    -0.65
    lived
    -0.65
    ynski
    -0.65
    chwitz
    -0.64
    uben
    -0.63
    bard
    -0.62
    ophen
    -0.61
    kept
    -0.60
    illusion
    -0.59
    uties
    -0.59
    POSITIVE LOGITS
    ingly
    1.02
     unsuspecting
    1.01
     electors
    0.93
     voters
    0.90
     him
    0.89
     gull
    0.88
     audiences
    0.87
     us
    0.87
     recal
    0.86
     passers
    0.84
    Activations Density 0.142%

    No Known Activations