© 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. 13617
    Prev
    Next
    INDEX
    Explanations

    descriptions of negative consequences or controversial situations

    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
    620
    +0.13
    0.6%
    283
    +0.11
    0.5%
    234
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    526
    +0.13
    -0.01
    620
    +0.11
    0.02
    351
    +0.10
    0.03
    Negative Logits
     Ank
    -0.65
     Recovery
    -0.60
    FORE
    -0.59
     Beginning
    -0.59
     Modified
    -0.59
     ROCK
    -0.58
    razil
    -0.57
    ————
    -0.57
     Earthquake
    -0.57
     Volunteers
    -0.56
    POSITIVE LOGITS
     nuance
    0.77
    ?).
    0.77
     sensibilities
    0.76
     euphem
    0.76
    .).
    0.75
     pund
    0.74
    entious
    0.72
     slightest
    0.72
     oneself
    0.71
     journalistic
    0.71
    Activations Density 1.015%

    No Known Activations