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

    Neuronpedia

    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. 9756
    Prev
    Next
    INDEX
    Explanations

    the word "call" in a context where it is being used to refer to a concept or action

    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
    191
    +0.12
    0.5%
    430
    +0.11
    0.5%
    131
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.12
    0.04
    131
    +0.11
    0.03
    603
    +0.10
    0.03
    Negative Logits
    isphere
    -0.79
     istg
    -0.72
    Written
    -0.67
    olitics
    -0.64
     mathemat
    -0.63
    ashington
    -0.62
    dayName
    -0.61
    oday
    -0.60
    ————
    -0.60
     Written
    -0.60
    POSITIVE LOGITS
    phas
    0.96
    igraph
    0.94
    backs
    0.92
     bullshit
    0.84
     attention
    0.81
     bluff
    0.81
    plates
    0.75
    ously
    0.74
    aghan
    0.72
     forth
    0.71
    Activations Density 0.049%

    No Known Activations