© 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 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. 43100
    Prev
    Next
    INDEX
    Explanations

    references to the highest levels or criteria in various contexts

    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
    496
    +0.10
    0.5%
    262
    +0.10
    0.5%
    187
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    262
    +0.10
    0.03
    194
    +0.10
    0.02
    153
    +0.09
    0.02
    Negative Logits
    izon
    -0.87
    vous
    -0.73
    pty
    -0.71
    neys
    -0.71
    md
    -0.70
     IMAGES
    -0.68
    prototype
    -0.66
    kay
    -0.66
    ajor
    -0.65
    fman
    -0.65
    POSITIVE LOGITS
     bidder
    1.15
     priority
    0.95
     rated
    0.95
     level
    0.93
     stakes
    0.90
     earners
    0.90
     priced
    0.89
     grade
    0.89
     concentrations
    0.88
     levels
    0.86
    Activations Density 0.017%

    No Known Activations