© 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. 750
    Prev
    Next
    INDEX
    Explanations

    mentions of planets in the context of explanations, discussions, or descriptions

    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
    288
    +0.10
    0.5%
    186
    +0.10
    0.5%
    414
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.02
    414
    +0.10
    0.02
    651
    +0.10
    0.02
    Negative Logits
    OUT
    -0.77
    DOS
    -0.76
    headers
    -0.75
    WER
    -0.72
    RANT
    -0.71
    EStreamFrame
    -0.69
    Ô
    -0.68
    IME
    -0.66
    Es
    -0.66
     Newsp
    -0.65
    POSITIVE LOGITS
    arium
    1.44
    ouched
    1.19
    oid
    1.11
     orbiting
    1.07
     Ceres
    0.99
     Venus
    0.97
     Earth
    0.94
    etary
    0.93
    oids
    0.92
    ogram
    0.92
    Activations Density 0.070%

    No Known Activations