© 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 12288
    9. 8-RES_FS12288-JB
    11. 1459
    Prev
    Next
    INDEX
    Explanations

    copyright information

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    +0.17
    0.8%
    296
    +0.10
    0.5%
    253
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.17
    -0.00
    733
    +0.10
    0.01
    587
    +0.10
    0.02
    Negative Logits
    externalActionCode
    -0.80
    trap
    -0.73
    sil
    -0.72
    worms
    -0.70
    sonian
    -0.70
    cos
    -0.69
    tra
    -0.68
    ·
    -0.68
    mental
    -0.67
    tail
    -0.67
    POSITIVE LOGITS
     reserved
    0.79
    ATIONS
    0.66
     Reserved
    0.64
     IMAGES
    0.64
     trademarks
    0.62
     âĶľâĶĢâĶĢ
    0.60
     Leban
    0.59
     Courtesy
    0.59
     Sanctuary
    0.58
     jurisdiction
    0.58
    Activations Density 0.069%

    No Known Activations