© 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. 784
    Prev
    Next
    INDEX
    Explanations

    references to specific locations or places

    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
    36
    +0.23
    1.1%
    481
    +0.18
    0.8%
    263
    +0.13
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    36
    +0.23
    0.13
    263
    +0.18
    0.07
    159
    +0.13
    0.06
    Negative Logits
    Own
    -0.81
    antry
    -0.80
    <?
    -0.77
    Arcade
    -0.76
    anism
    -0.74
    Hon
    -0.71
    AIDS
    -0.69
    via
    -0.68
    agree
    -0.68
    CI
    -0.68
    POSITIVE LOGITS
     handful
    1.09
     lot
    1.05
     couple
    0.98
     few
    0.96
     combination
    0.93
     spokesperson
    0.93
     colleague
    0.92
     majority
    0.92
     concerted
    0.91
     woman
    0.91
    Activations Density 0.274%

    No Known Activations