Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 12288
    5. 8-RES_FS12288-JB
    6. 214
    Prev
    Next
    INDEX
    Explanations

    words related to cyber security and digital attacks

    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
    447
    +0.17
    0.8%
    468
    +0.10
    0.5%
    507
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    468
    +0.17
    0.05
    271
    +0.10
    0.05
    59
    +0.10
    0.05
    Negative Logits
    mad
    -0.78
    href
    -0.78
    nings
    -0.75
    trap
    -0.74
    fare
    -0.73
    ciplinary
    -0.72
    edition
    -0.72
    iac
    -0.72
     spons
    -0.72
    deal
    -0.70
    POSITIVE LOGITS
     them
    0.93
     him
    0.91
     unsuspecting
    0.76
     Them
    0.75
     the
    0.73
     us
    0.69
     someone
    0.69
     Buckingham
    0.68
     THEM
    0.68
     anybody
    0.68
    Activations Density 0.451%

    No Known Activations