© 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. 1175
    Prev
    Next
    INDEX
    Explanations

    terms related to natural disasters and infrastructure damages, particularly those caused by floods

    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
    419
    +0.12
    0.5%
    129
    +0.11
    0.5%
    178
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    419
    +0.12
    0.03
    129
    +0.11
    0.03
    288
    +0.10
    0.03
    Negative Logits
    ccording
    -0.84
    Ģ
    -0.73
    ooth
    -0.73
    cius
    -0.72
    CRIPTION
    -0.69
    ACP
    -0.67
    atel
    -0.65
    doms
    -0.65
    AAAAAAAA
    -0.65
     Lifetime
    -0.65
    POSITIVE LOGITS
    waters
    1.24
    plain
    1.00
     Flood
    0.93
     Katrina
    0.91
     waters
    0.87
     plains
    0.87
    lights
    0.86
     inund
    0.86
    water
    0.84
    lighting
    0.83
    Activations Density 0.085%

    No Known Activations