© 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. 14294
    Prev
    Next
    INDEX
    Explanations

    explanations or clarifications in a text

    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
    679
    +0.15
    0.7%
    195
    +0.12
    0.5%
    81
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    195
    +0.15
    0.02
    679
    +0.12
    0.03
    626
    +0.11
    0.02
    Negative Logits
    luster
    -0.72
     Sabha
    -0.71
    oreal
    -0.69
    estial
    -0.68
    ches
    -0.67
    Ranked
    -0.66
    sembly
    -0.66
    illet
    -0.65
    etooth
    -0.63
    bour
    -0.63
    POSITIVE LOGITS
     why
    1.30
     WHY
    1.11
     how
    1.02
    why
    0.97
    how
    0.84
     succinct
    0.83
     concise
    0.82
     HOW
    0.76
     plainly
    0.73
    ļéĨĴ
    0.73
    Activations Density 0.037%

    No Known Activations