© 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. 10102
    Prev
    Next
    INDEX
    Explanations

    natural disaster-related terms, focusing on storms and their impact

    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
    379
    +0.12
    0.6%
    116
    +0.10
    0.5%
    361
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    379
    +0.12
    0.03
    419
    +0.10
    0.03
    326
    +0.10
    0.02
    Negative Logits
    buquerque
    -0.88
    atem
    -0.84
    ADRA
    -0.84
    olars
    -0.82
    cius
    -0.80
    sembly
    -0.79
    ublic
    -0.79
    atown
    -0.78
    unal
    -0.78
    iture
    -0.76
    POSITIVE LOGITS
    tro
    1.20
     surge
    1.01
     storms
    1.00
    storm
    0.97
     Sandy
    0.95
    storms
    0.95
     storm
    0.90
     Irma
    0.89
     clouds
    0.83
     surges
    0.82
    Activations Density 0.057%

    No Known Activations