© 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 24576
    9. 8-RES_FS24576-JB
    11. 18956
    Prev
    Next
    INDEX
    Explanations

    words related to military combat and warfare

    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
    288
    +0.19
    0.9%
    85
    +0.09
    0.4%
    16
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.19
    0.02
    85
    +0.09
    0.01
    16
    +0.09
    0.01
    Negative Logits
    gow
    -0.73
    ħĭ
    -0.72
    eus
    -0.72
     Seym
    -0.66
    ocl
    -0.65
    bye
    -0.63
    sonian
    -0.62
    ocument
    -0.61
     Boh
    -0.61
     explan
    -0.60
    POSITIVE LOGITS
    iveness
    1.13
    fighting
    0.94
    ant
    0.93
    ting
    0.92
    ants
    0.92
     fatig
    0.84
    ments
    0.83
    fighter
    0.80
    fighters
    0.80
     prowess
    0.79
    Activations Density 0.020%

    No Known Activations