© 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. 18200
    Prev
    Next
    INDEX
    Explanations

    mentions of political and social issues, particularly related to government policies and societal opinions

    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
    481
    +0.18
    0.9%
    139
    +0.13
    0.6%
    278
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    368
    +0.18
    0.05
    139
    +0.13
    0.05
    561
    +0.11
    0.05
    Negative Logits
    OUGH
    -0.67
     Zimmer
    -0.65
    enthal
    -0.63
    HAM
    -0.63
     Stephenson
    -0.63
    clusive
    -0.63
    Marco
    -0.63
    Actor
    -0.62
    STON
    -0.61
     Compton
    -0.61
    POSITIVE LOGITS
     behaved
    0.80
     chose
    0.79
     cannot
    0.78
     withdrew
    0.76
     deserve
    0.76
     were
    0.76
     refuse
    0.75
     are
    0.75
     have
    0.75
     disclaim
    0.75
    Activations Density 0.192%

    No Known Activations