© 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. 4448
    Prev
    Next
    INDEX
    Explanations

    positive traits or qualities related to a particular topic or subject, such as reviews or descriptions

    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
    480
    +0.48
    2.5%
    266
    +0.11
    0.6%
    49
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    480
    +0.48
    0.00
    690
    +0.11
    0.05
    655
    +0.08
    0.05
    Negative Logits
    iries
    -0.62
    agents
    -0.61
     roofs
    -0.59
    safety
    -0.58
    afety
    -0.58
     neighbours
    -0.58
    anwhile
    -0.58
    volent
    -0.57
    forces
    -0.57
    iencies
    -0.57
    POSITIVE LOGITS
     autobi
    0.89
     parody
    0.86
     sequel
    0.86
     masterpiece
    0.83
     enjoyable
    0.80
     satire
    0.80
     titled
    0.80
     plagiar
    0.78
     chronological
    0.77
     cinematic
    0.77
    Activations Density 0.747%

    No Known Activations