© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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 12288
    9. 8-RES_FS12288-JB
    11. 9422
    Prev
    Next
    INDEX
    Explanations

    words related to crimes, law enforcement, and political controversy

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    16
    +0.13
    0.6%
    354
    +0.10
    0.5%
    762
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    16
    +0.13
    0.03
    371
    +0.10
    0.02
    357
    +0.09
    0.02
    Negative Logits
    æ©Ł
    -0.71
    pires
    -0.67
     Philos
    -0.64
    UTH
    -0.61
     Warm
    -0.61
    ipation
    -0.61
    ĨĴ
    -0.60
    otive
    -0.57
    å¸
    -0.57
    inent
    -0.57
    POSITIVE LOGITS
    ched
    1.17
    chens
    1.12
    ches
    1.08
    achi
    1.03
    ters
    1.00
    boxes
    0.93
    tle
    0.91
    box
    0.90
    ting
    0.84
    dos
    0.82
    Activations Density 0.072%

    No Known Activations