© 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. 8099
    Prev
    Next
    INDEX
    Explanations

    uncommon or unique phrases that involve the idea of "going out of one's way" or taking extra steps in some context

    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
    373
    +0.20
    1.0%
    297
    +0.12
    0.5%
    292
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.20
    0.05
    297
    +0.12
    0.04
    50
    +0.10
    0.05
    Negative Logits
    Exit
    -0.63
     Anyway
    -0.62
    inventoryQuantity
    -0.62
     Disapp
    -0.62
     Personality
    -0.56
     tumble
    -0.55
     outweigh
    -0.55
     Amen
    -0.54
     ailments
    -0.53
    exist
    -0.53
    POSITIVE LOGITS
     instruct
    0.71
    lander
    0.69
     solic
    0.67
    fold
    0.65
     incentiv
    0.65
     persu
    0.65
     ta
    0.64
     exh
    0.63
     persuaded
    0.62
    İ
    0.61
    Activations Density 0.371%

    No Known Activations