© 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. 16544
    Prev
    Next
    INDEX
    Explanations

    information related to a specific medical condition or program named "PD"

    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
    453
    +0.13
    0.6%
    456
    +0.11
    0.5%
    742
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    453
    +0.13
    0.01
    348
    +0.11
    0.01
    116
    +0.11
    0.01
    Negative Logits
    ppo
    -0.85
    bourg
    -0.80
    natureconservancy
    -0.72
    ãĤ¤ãĥĪ
    -0.71
    ansas
    -0.70
    itud
    -0.68
     âĶľ
    -0.67
    fram
    -0.67
    hold
    -0.66
    bestos
    -0.65
    POSITIVE LOGITS
    iamond
    0.97
    GF
    0.89
    ouble
    0.87
    ATES
    0.81
    DR
    0.81
    ragon
    0.80
    ynasty
    0.80
    iants
    0.78
    illon
    0.78
    iaries
    0.78
    Activations Density 0.019%

    No Known Activations