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

    Neuronpedia

    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. 24246
    Prev
    Next
    INDEX
    Explanations

    phrases related to belief and conviction

    oai_token-act-pair · gpt-4o-miniTriggered 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
    87
    +0.55
    3.6%
    314
    +0.12
    0.7%
    756
    +0.10
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    675
    +0.55
    0.16
    37
    +0.12
    0.13
    314
    +0.10
    0.04
    Negative Logits
     vulner
    -0.51
     sacrific
    -0.50
     pyramid
    -0.48
     pigeon
    -0.45
     convenience
    -0.44
     scattering
    -0.43
     surv
    -0.43
     silhou
    -0.43
     seiz
    -0.42
     nude
    -0.42
    POSITIVE LOGITS
    ï¸ı
    0.87
    east
    0.65
    said
    0.64
    tre
    0.62
    shall
    0.61
    span
    0.61
    ski
    0.61
    _>
    0.60
    then
    0.60
    should
    0.60
    Activations Density 0.511%

    No Known Activations