© 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. 4366
    Prev
    Next
    INDEX
    Explanations

    the word "Dig"

    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
    211
    +0.12
    0.6%
    569
    +0.12
    0.5%
    140
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    140
    +0.12
    0.01
    211
    +0.12
    0.01
    569
    +0.10
    0.01
    Negative Logits
     Mara
    -0.69
     ACTIONS
    -0.67
     wart
    -0.66
     Rae
    -0.63
    Reloaded
    -0.63
     Fenrir
    -0.63
    OHN
    -0.60
     Karin
    -0.59
     Archangel
    -0.59
     SAL
    -0.57
    POSITIVE LOGITS
    gers
    1.35
    ging
    1.25
    raph
    1.23
    itized
    1.21
    imon
    1.20
    ression
    1.19
    itial
    1.14
    itals
    1.12
    ibility
    1.11
    ger
    1.09
    Activations Density 0.011%

    No Known Activations