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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 2904
    Prev
    Next
    INDEX
    Explanations

    articles and research studies published in the journal "Nature" and "Nature Communications."

    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
    +0.12
    0.5%
    269
    +0.12
    0.5%
    271
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    271
    +0.12
    0.02
    631
    +0.12
    0.02
    269
    +0.10
    0.02
    Negative Logits
    haul
    -0.82
     perjury
    -0.71
     ceilings
    -0.71
     maid
    -0.70
     piping
    -0.68
     awoken
    -0.66
     deport
    -0.64
     confisc
    -0.64
    landish
    -0.63
     leveled
    -0.63
    POSITIVE LOGITS
     Communications
    1.07
     Psychiatry
    1.05
     Publishing
    0.99
     Genetics
    0.99
     Materials
    0.99
     Communication
    0.98
     Medicine
    0.95
     Proceedings
    0.95
     Methods
    0.94
     Letters
    0.93
    Activations Density 0.034%

    No Known Activations