© 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. 8587
    Prev
    Next
    INDEX
    Explanations

    specific mentions of prepositions followed by certain nouns

    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
    269
    +0.11
    0.5%
    649
    +0.09
    0.4%
    707
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    269
    +0.11
    0.04
    654
    +0.09
    0.04
    471
    +0.08
    0.04
    Negative Logits
    ν
    -0.65
    ̶
    -0.62
    antics
    -0.61
    wig
    -0.61
    amus
    -0.60
    ruciating
    -0.60
    ÏĤ
    -0.59
     wrench
    -0.57
    vor
    -0.57
    inos
    -0.57
    POSITIVE LOGITS
    screen
    1.19
     behalf
    1.10
    erous
    1.04
    site
    1.02
     billboards
    0.99
    ibaba
    0.93
     Pastebin
    0.92
    sets
    0.90
    shore
    0.89
     eBay
    0.89
    Activations Density 0.145%

    No Known Activations