© 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. 10299
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal concepts

    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
    480
    +0.35
    1.7%
    351
    +0.13
    0.7%
    393
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    351
    +0.35
    0.08
    0
    +0.13
    0.09
    484
    +0.11
    0.09
    Negative Logits
    DX
    -0.66
    meric
    -0.63
    Voice
    -0.63
    Morning
    -0.62
    atan
    -0.62
    igger
    -0.61
     Powered
    -0.60
    owered
    -0.60
    ãģĨ
    -0.60
    uct
    -0.60
    POSITIVE LOGITS
     thereby
    1.43
     consequently
    1.37
     hence
    1.34
     thus
    1.32
     therefore
    1.20
     secondly
    1.20
     furthermore
    1.06
     vice
    1.04
     subsequently
    0.92
     thereafter
    0.91
    Activations Density 0.542%

    No Known Activations