© 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 24576
    9. 8-RES_FS24576-JB
    11. 10786
    Prev
    Next
    INDEX
    Explanations

    information related to projects, activities, and regulations, with a focus on Cuban travel and community rankings

    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
    76
    +0.11
    0.5%
    108
    +0.11
    0.5%
    6
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    76
    +0.11
    0.04
    145
    +0.11
    0.04
    232
    +0.10
    0.03
    Negative Logits
     reluct
    -0.89
    conservancy
    -0.88
    .ãĢį
    -0.86
     tremend
    -0.79
     corrid
    -0.79
    uders
    -0.75
    rising
    -0.75
    ÂŃ
    -0.74
    picture
    -0.74
    pill
    -0.73
    POSITIVE LOGITS
     Miscellaneous
    1.20
     Various
    1.10
     âĵĺ
    1.06
     ----------------------------------------------------------------
    0.93
     Other
    0.91
     Total
    0.88
     None
    0.87
     Others
    0.86
     Submit
    0.86
     Thousands
    0.85
    Activations Density 0.171%

    No Known Activations