© 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. 682
    Prev
    Next
    INDEX
    Explanations

    Proper names and specialized terms related to historical events or people

    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
    481
    +0.22
    1.0%
    447
    +0.22
    1.0%
    232
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.22
    0.23
    659
    +0.22
    0.19
    667
    +0.09
    0.13
    Negative Logits
     RELE
    -0.73
     acron
    -0.72
     ASC
    -0.69
     INTER
    -0.67
     IMP
    -0.66
     SAN
    -0.66
     iP
    -0.65
     "\
    -0.63
     Princ
    -0.62
     nic
    -0.61
    POSITIVE LOGITS
    inar
    0.87
    idy
    0.85
    rake
    0.82
    zon
    0.81
    ormon
    0.81
    itan
    0.80
    ipal
    0.80
    ilus
    0.80
    ocity
    0.80
    eez
    0.80
    Activations Density 3.912%

    No Known Activations