© 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. 14681
    Prev
    Next
    INDEX
    Explanations

    Arabic, Persian, Kurdish, Marathi, Hindi, Icelandic, Armenian, Urdu language phrases

    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
    679
    +0.21
    1.0%
    199
    +0.10
    0.5%
    392
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.21
    0.02
    365
    +0.10
    0.01
    392
    +0.09
    0.02
    Negative Logits
    phrine
    -0.78
    20439
    -0.76
    issance
    -0.72
    iber
    -0.70
    ograp
    -0.69
    âĸ¬âĸ¬
    -0.68
    ected
    -0.67
    aching
    -0.67
    ollar
    -0.65
     Saunders
    -0.64
    POSITIVE LOGITS
    atoon
    0.81
    edIn
    0.80
    aine
    0.77
    ozy
    0.75
    ر
    0.72
    score
    0.71
    sburgh
    0.70
    rats
    0.69
    s
    0.68
    anasia
    0.68
    Activations Density 0.037%

    No Known Activations