© 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. 15745
    Prev
    Next
    INDEX
    Explanations

    the name "Vicente."

    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
    659
    +0.14
    0.6%
    231
    +0.11
    0.5%
    415
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.14
    0.01
    538
    +0.11
    0.01
    231
    +0.10
    0.01
    Negative Logits
    Ô
    -0.90
    */(
    -0.84
    mble
    -0.76
    xual
    -0.73
    tenance
    -0.71
    stretched
    -0.70
    CLASS
    -0.69
     EntityItem
    -0.67
    landish
    -0.65
    dress
    -0.64
    POSITIVE LOGITS
    arious
    1.41
    inity
    1.12
    ente
    1.11
    isions
    1.07
    entric
    1.05
    abulary
    1.01
    uous
    1.00
    iple
    0.99
    eland
    0.97
    imates
    0.97
    Activations Density 0.022%

    No Known Activations