© 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. 5498
    Prev
    Next
    INDEX
    Explanations

    entities or objects that are the focus of attention in a project or discussion

    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
    328
    +0.12
    0.5%
    194
    +0.11
    0.5%
    296
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    296
    +0.12
    0.02
    5
    +0.11
    0.02
    671
    +0.10
    0.02
    Negative Logits
    sports
    -0.84
    ancies
    -0.80
    doms
    -0.76
    contin
    -0.73
    sung
    -0.69
    fried
    -0.66
    etheless
    -0.65
    rams
    -0.65
    nces
    -0.64
    cats
    -0.64
    POSITIVE LOGITS
     checkout
    0.77
     introdu
    0.77
     reaction
    0.71
     foremost
    0.71
     responders
    0.70
     ali
    0.69
     greeted
    0.68
     blush
    0.67
     knocks
    0.67
     glance
    0.66
    Activations Density 0.098%

    No Known Activations