© 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 49152
    9. 8-RES_FS49152-JB
    11. 37472
    Prev
    Next
    INDEX
    Explanations

    commands related to file and package management in a command-line interface

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    700
    +0.13
    0.6%
    373
    +0.12
    0.5%
    165
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.13
    0.01
    700
    +0.12
    0.01
    345
    +0.10
    0.01
    Negative Logits
    ourgeois
    -0.73
    isner
    -0.68
    externalActionCode
    -0.67
    BRE
    -0.67
    ingers
    -0.67
    emi
    -0.63
    ricks
    -0.63
    icz
    -0.63
    ön
    -0.63
     gearing
    -0.62
    POSITIVE LOGITS
     ~/.
    0.76
     "$
    0.76
     [-
    0.73
     "%
    0.73
    lift
    0.73
    dir
    0.72
     ~/
    0.70
     '/
    0.69
     (-
    0.67
     xp
    0.66
    Activations Density 0.014%

    No Known Activations