© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 2250
    Prev
    Next
    INDEX
    Explanations

    punctuation marks within words

    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
    314
    +0.15
    0.7%
    152
    +0.11
    0.5%
    377
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    102
    +0.15
    0.04
    152
    +0.11
    0.04
    205
    +0.10
    0.03
    Negative Logits
     Archdemon
    -0.80
     Hendricks
    -0.70
     printer
    -0.69
     Burnett
    -0.69
     Ames
    -0.68
     polyg
    -0.65
     collabor
    -0.65
     dispatch
    -0.64
     printers
    -0.64
     Argon
    -0.64
    POSITIVE LOGITS
    cause
    1.61
    til
    1.45
    tis
    1.33
    Interstitial
    1.33
    Mech
    1.28
    taboola
    1.23
    Cause
    1.21
    round
    1.10
    MpServer
    1.10
    Mania
    1.05
    Activations Density 0.039%

    No Known Activations