© 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 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. 18282
    Prev
    Next
    INDEX
    Explanations

    references to specific names or terms, like "Levi" and "Allahu Akbar," connected to individuals or events

    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
    308
    +0.13
    0.6%
    566
    +0.12
    0.5%
    195
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.13
    0.02
    343
    +0.12
    0.02
    659
    +0.10
    0.02
    Negative Logits
    lain
    -1.01
    erick
    -0.96
    es
    -0.87
    maid
    -0.83
    hips
    -0.81
    grave
    -0.79
    rification
    -0.79
    esian
    -0.78
    ification
    -0.78
    lie
    -0.76
    POSITIVE LOGITS
     Strauss
    0.92
    ptions
    0.90
    ×IJ
    0.82
    verages
    0.80
    Wan
    0.79
    resent
    0.78
    asca
    0.75
    ught
    0.75
    abetic
    0.73
    vention
    0.68
    Activations Density 0.042%

    No Known Activations