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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 19921
    Prev
    Next
    INDEX
    Explanations

    computer code snippets and commands

    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
    365
    +0.13
    0.6%
    388
    +0.11
    0.5%
    566
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    365
    +0.13
    0.02
    659
    +0.11
    0.03
    289
    +0.10
    0.01
    Negative Logits
    EVA
    -0.71
    heit
    -0.65
     Dresden
    -0.64
    enegger
    -0.63
    SHIP
    -0.61
    packing
    -0.60
    士
    -0.60
     Lesbian
    -0.60
     Luk
    -0.59
     Labrador
    -0.59
    POSITIVE LOGITS
    entimes
    0.93
    isoft
    0.92
    mble
    0.92
    cean
    0.89
    umen
    0.88
    yssey
    0.85
    ertain
    0.82
    anium
    0.78
    gment
    0.77
    aml
    0.76
    Activations Density 0.098%

    No Known Activations