© 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
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 12288
    5. 8-RES_FS12288-JB
    6. 736
    Prev
    Next
    INDEX
    Explanations

    the word "try" followed by single digit numbers or specific phrases

    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
    641
    +0.10
    0.4%
    202
    +0.10
    0.4%
    665
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.10
    0.05
    351
    +0.10
    0.04
    665
    +0.09
    0.04
    Negative Logits
    head
    -0.70
    dylib
    -0.65
    HEAD
    -0.65
    effect
    -0.62
    resent
    -0.62
    edly
    -0.61
    atform
    -0.60
    goers
    -0.59
    Construction
    -0.59
    inational
    -0.59
    POSITIVE LOGITS
     unsuccessfully
    1.12
    nir
    0.87
     harder
    0.85
    anke
    0.75
     contacting
    0.71
    irtual
    0.68
     desperately
    0.66
    outs
    0.66
    bles
    0.64
     valiant
    0.64
    Activations Density 0.062%

    No Known Activations