© 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. 10966
    Prev
    Next
    INDEX
    Explanations

    terms related to technology and development, such as "technical goal", "code smells", and "startup ecosystem"

    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
    640
    +0.24
    1.1%
    373
    +0.13
    0.6%
    447
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    640
    +0.24
    0.08
    179
    +0.13
    0.05
    615
    +0.11
    0.04
    Negative Logits
    hement
    -0.90
    feed
    -0.64
    lected
    -0.61
    tered
    -0.60
    finger
    -0.59
    в
    -0.59
    ean
    -0.57
    going
    -0.57
    ighth
    -0.56
    quist
    -0.56
    POSITIVE LOGITS
    ccording
    0.94
    ³³³³³³³³³³³³³³³³
    0.82
    ³³³³³³³³
    0.75
    ³³³
    0.72
     Throughout
    0.68
    ³³³³
    0.67
     Initially
    0.65
     Similar
    0.64
     Despite
    0.64
     Contrary
    0.62
    Activations Density 0.138%

    No Known Activations