© 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. 7525
    Prev
    Next
    INDEX
    Explanations

    phrases related to discussions or explanations about various topics

    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
    286
    +0.11
    0.5%
    206
    +0.10
    0.5%
    199
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    206
    +0.11
    0.01
    123
    +0.10
    0.01
    665
    +0.10
    0.01
    Negative Logits
    issy
    -0.67
    hatt
    -0.65
     rooft
    -0.64
     apples
    -0.63
    wana
    -0.62
     Owens
    -0.60
     hog
    -0.59
    urden
    -0.58
     Reuters
    -0.58
    unker
    -0.58
    POSITIVE LOGITS
     havoc
    0.98
     revolutions
    0.88
    uate
    0.82
     irreversible
    0.73
    dL
    0.71
     pandemonium
    0.71
     unforeseen
    0.70
    ounter
    0.70
     alterations
    0.69
    versible
    0.69
    Activations Density 0.046%

    No Known Activations