© 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. 12389
    Prev
    Next
    INDEX
    Explanations

    terms related to scientific research and analysis, as well as some references to physical locations or figures

    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
    311
    +0.10
    0.4%
    352
    +0.09
    0.4%
    389
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.04
    763
    +0.09
    0.03
    311
    +0.08
    0.01
    Negative Logits
    PDATE
    -0.80
     Penet
    -0.76
     Doctrine
    -0.73
    WARD
    -0.67
     Belt
    -0.67
    beard
    -0.67
    theless
    -0.66
     LORD
    -0.65
    spin
    -0.62
     rule
    -0.61
    POSITIVE LOGITS
    ators
    2.04
    ations
    1.99
    ator
    1.83
    ative
    1.81
    atory
    1.77
    ating
    1.75
    ational
    1.73
    ates
    1.63
    atives
    1.56
    atively
    1.50
    Activations Density 0.067%

    No Known Activations