© 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. 22407
    Prev
    Next
    INDEX
    Explanations

    descriptors related to ability and proficiency, especially in intellectual or skill-based 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
    481
    +0.11
    0.5%
    447
    +0.10
    0.5%
    46
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    435
    +0.11
    0.04
    665
    +0.10
    0.05
    724
    +0.10
    0.05
    Negative Logits
     unwanted
    -0.65
    JP
    -0.65
    reddits
    -0.59
     Suffolk
    -0.58
     Carbuncle
    -0.57
    ãĤ¼
    -0.57
    Welcome
    -0.55
     fond
    -0.55
    NetMessage
    -0.55
    Page
    -0.53
    POSITIVE LOGITS
     ASAP
    1.12
     Otherwise
    1.08
     lest
    1.03
     preferably
    0.90
     urgently
    0.86
     manually
    0.81
    Otherwise
    0.76
     beforehand
    0.75
     BEFORE
    0.75
     Pwr
    0.75
    Activations Density 0.523%

    No Known Activations