© 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 49152
    9. 8-RES_FS49152-JB
    11. 45458
    Prev
    Next
    INDEX
    Explanations

    instances of citations or references within the text

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    640
    +0.19
    0.9%
    14
    +0.09
    0.4%
    762
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    640
    +0.19
    0.04
    160
    +0.09
    0.03
    6
    +0.09
    0.03
    Negative Logits
     endeavour
    -0.81
     steroids
    -0.77
     colours
    -0.73
     advoc
    -0.72
     organis
    -0.70
    izont
    -0.70
     canvas
    -0.69
     practise
    -0.69
     endeav
    -0.68
     clay
    -0.68
    POSITIVE LOGITS
    Sources
    1.01
    According
    0.96
    WASHINGTON
    0.95
    Officials
    0.95
     largeDownload
    0.92
    Authorities
    0.92
    Prosecutors
    0.89
    SPONSORED
    0.88
    BRE
    0.88
    Parts
    0.87
    Activations Density 0.085%

    No Known Activations