© 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. 17693
    Prev
    Next
    INDEX
    Explanations

    government-related entities or programs, medical care services, and civil or human rights contexts

    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
    264
    +0.09
    0.5%
    28
    +0.09
    0.4%
    407
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.09
    0.04
    365
    +0.09
    0.02
    396
    +0.09
    0.03
    Negative Logits
     referen
    -0.66
     fastball
    -0.66
    milo
    -0.65
    ħĭ
    -0.63
    代
    -0.61
     mug
    -0.61
     sticker
    -0.58
     heck
    -0.58
     unch
    -0.56
     swast
    -0.54
    POSITIVE LOGITS
    uty
    0.66
     Affairs
    0.66
     Services
    0.65
    esson
    0.63
    ļ
    0.62
    utics
    0.61
     alike
    0.61
     Products
    0.61
    giene
    0.61
     Depend
    0.58
    Activations Density 0.140%

    No Known Activations