© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 17392
    Prev
    Next
    INDEX
    Explanations

    action-related words pointing to instructions or advice for different situations

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    447
    +0.14
    0.6%
    334
    +0.11
    0.5%
    416
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    184
    +0.14
    0.03
    604
    +0.11
    0.03
    738
    +0.11
    0.03
    Negative Logits
    ](
    -0.70
    hell
    -0.62
    edition
    -0.61
    alist
    -0.60
     territ
    -0.59
    adal
    -0.59
    de
    -0.59
    ctrl
    -0.58
    ocaust
    -0.58
    emale
    -0.58
    POSITIVE LOGITS
     yourself
    1.07
    cknow
    1.00
     yourselves
    0.97
     Yourself
    0.97
     your
    0.83
     carefully
    0.82
    ²¾
    0.78
     preferably
    0.75
     Your
    0.73
     sparing
    0.73
    Activations Density 0.164%

    No Known Activations