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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 73
    Prev
    Next
    INDEX
    Explanations

    phrases related to criminal investigations and incidents

    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
    566
    +0.11
    0.5%
    171
    +0.10
    0.4%
    564
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    566
    +0.11
    0.06
    86
    +0.10
    0.04
    551
    +0.10
    0.04
    Negative Logits
    opian
    -0.74
    die
    -0.71
    etus
    -0.71
     Virtue
    -0.70
    âķ
    -0.69
     congress
    -0.67
    sed
    -0.66
     Spir
    -0.66
     proportions
    -0.64
    Creat
    -0.64
    POSITIVE LOGITS
     investigating
    1.28
     investigate
    1.01
     dispatched
    1.00
    intendent
    0.98
     notified
    0.95
     uncover
    0.95
     searching
    0.94
     alerted
    0.94
     patrolling
    0.94
    nsic
    0.92
    Activations Density 0.229%

    No Known Activations