© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 26584
    Prev
    Next
    INDEX
    Explanations

    numerical specifications related to hardware components

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    495
    +0.10
    0.5%
    326
    +0.10
    0.5%
    468
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.10
    0.02
    495
    +0.10
    0.02
    252
    +0.10
    0.02
    Negative Logits
    DonaldTrump
    -0.68
    matically
    -0.68
    emort
    -0.65
    artisan
    -0.61
    htaking
    -0.60
    ibilities
    -0.59
    hoe
    -0.58
    ographers
    -0.58
    aneers
    -0.56
    undo
    -0.56
    POSITIVE LOGITS
     respectively
    0.97
    kHz
    0.91
     ratio
    0.79
    )=(
    0.78
     istg
    0.78
     kHz
    0.77
    MHz
    0.75
    /
    0.75
    /?
    0.74
     Ratio
    0.74
    Activations Density 0.026%

    No Known Activations