© 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 12288
    9. 8-RES_FS12288-JB
    11. 9802
    Prev
    Next
    INDEX
    Explanations

    references to current events and societal issues

    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
    276
    +0.12
    0.5%
    201
    +0.11
    0.5%
    226
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    279
    +0.12
    0.03
    274
    +0.11
    0.03
    276
    +0.10
    0.03
    Negative Logits
    »Ĵ
    -0.70
     Antar
    -0.68
    emy
    -0.67
    geons
    -0.67
    Reader
    -0.64
    oru
    -0.63
    ename
    -0.63
    idav
    -0.61
    ewitness
    -0.61
    eday
    -0.60
    POSITIVE LOGITS
     ado
    0.96
     transpired
    0.96
     happened
    0.88
     happening
    0.86
     nonsense
    0.84
    together
    0.82
    stuff
    0.81
     madness
    0.81
     fuss
    0.81
     stuff
    0.77
    Activations Density 0.118%

    No Known Activations