© 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 24576
    9. 8-RES_FS24576-JB
    11. 18880
    Prev
    Next
    INDEX
    Explanations

    comparisons or degrees of something, particularly with the word "than"

    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
    373
    +0.19
    0.9%
    655
    +0.14
    0.7%
    183
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    655
    +0.19
    0.04
    183
    +0.14
    0.03
    432
    +0.11
    0.03
    Negative Logits
    ilic
    -0.79
    DH
    -0.74
    ICAN
    -0.69
    ç·
    -0.69
    ffen
    -0.67
    ranch
    -0.67
    iless
    -0.66
    ensis
    -0.64
    Express
    -0.64
    Interstitial
    -0.64
    POSITIVE LOGITS
     doubling
    0.85
    whelming
    0.81
    atos
    0.76
     tripled
    0.75
     half
    0.74
    etheless
    0.73
     8000
    0.73
     80
    0.73
     doubled
    0.73
     700
    0.70
    Activations Density 0.027%

    No Known Activations