© 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 768
    9. 8-RES_FS768-JB
    11. 700
    Prev
    Next
    INDEX
    Explanations

    Jewish religious figures and related locations or events

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    574
    +0.11
    0.5%
    453
    +0.10
    0.4%
    468
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.11
    0.11
    574
    +0.10
    0.08
    326
    +0.10
    0.12
    Negative Logits
     Mandela
    -0.69
    creen
    -0.68
     Corpus
    -0.62
     Schumer
    -0.62
    rium
    -0.61
    terson
    -0.61
     SOS
    -0.61
    gom
    -0.61
     Franç
    -0.59
     Granger
    -0.59
    POSITIVE LOGITS
    senal
    1.32
    outine
    1.18
    haps
    1.18
    ansom
    1.16
    ascal
    1.13
    ework
    1.09
    itchie
    1.08
    ussia
    1.08
    abbit
    1.07
    agnar
    1.01
    Activations Density 3.878%

    No Known Activations