© 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. 16917
    Prev
    Next
    INDEX
    Explanations

    comparisons between different items or concepts

    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
    255
    +0.11
    0.5%
    679
    +0.10
    0.5%
    94
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    131
    +0.11
    0.02
    438
    +0.10
    0.02
    94
    +0.10
    0.02
    Negative Logits
    authorized
    -0.76
     Permanent
    -0.67
     Publication
    -0.67
    Bay
    -0.64
    kamp
    -0.58
     recess
    -0.57
    isle
    -0.57
    STAR
    -0.57
    utters
    -0.57
    FL
    -0.57
    POSITIVE LOGITS
     favorably
    0.99
    lihood
    0.79
     compare
    0.79
     comparing
    0.77
     comparison
    0.77
     Compare
    0.76
     thereto
    0.74
    Compare
    0.72
     athlet
    0.71
     compared
    0.69
    Activations Density 0.025%

    No Known Activations