© 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. 860
    Prev
    Next
    INDEX
    Explanations

    language related to decision-making and legal contexts

    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
    762
    +0.11
    0.5%
    193
    +0.10
    0.5%
    715
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    401
    +0.11
    0.10
    715
    +0.10
    0.08
    762
    +0.10
    0.09
    Negative Logits
    abilia
    -0.71
    actionDate
    -0.69
    ãĤ¯
    -0.66
    col
    -0.65
    tags
    -0.63
    classic
    -0.63
    piring
    -0.62
    Minecraft
    -0.61
    taining
    -0.61
     Tropical
    -0.61
    POSITIVE LOGITS
     cannot
    1.02
     spends
    0.99
     behaved
    0.97
     knows
    0.96
     realizes
    0.96
     lacks
    0.95
     prefers
    0.95
     actively
    0.95
     chooses
    0.94
     ought
    0.94
    Activations Density 0.693%

    No Known Activations