© 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. 22426
    Prev
    Next
    INDEX
    Explanations

    proper nouns, particularly names of people and teams in sports contexts

    oai_token-act-pair · gpt-4o-miniTriggered 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
    447
    +0.21
    1.0%
    481
    +0.13
    0.6%
    430
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.21
    0.06
    430
    +0.13
    0.07
    224
    +0.12
    0.06
    Negative Logits
    higher
    -0.62
    ãĤ¤ãĥĪ
    -0.61
    ãĥ¬
    -0.61
    κ
    -0.60
     ..."
    -0.59
    ãĤ¤
    -0.58
    appropriate
    -0.58
    trop
    -0.58
    Scar
    -0.58
    ãĥķãĤ¡
    -0.57
    POSITIVE LOGITS
     also
    1.05
     testified
    1.04
    enegger
    1.02
     meanwhile
    1.01
     insists
    1.00
     admits
    1.00
     reportedly
    0.99
     survived
    0.97
     acknowledges
    0.97
     told
    0.95
    Activations Density 0.182%

    No Known Activations