© 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. 9745
    Prev
    Next
    INDEX
    Explanations

    instances of people staring intently or engaging in intense gazes

    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
    161
    +0.12
    0.5%
    125
    +0.09
    0.4%
    407
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    161
    +0.12
    0.03
    125
    +0.09
    0.03
    435
    +0.09
    0.03
    Negative Logits
     orchestr
    -0.67
    ctr
    -0.65
    cially
    -0.63
    Fact
    -0.62
     graft
    -0.62
    erto
    -0.62
    Claim
    -0.61
    UFF
    -0.60
    cial
    -0.60
    ãĤ¡
    -0.59
    POSITIVE LOGITS
     gaze
    1.13
     stare
    1.07
     blank
    1.01
     stares
    1.00
     glare
    0.98
     blinking
    0.94
     challeng
    0.90
     staring
    0.90
     smir
    0.89
     curiously
    0.88
    Activations Density 0.348%

    No Known Activations