© 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 EvalsExportsinterp-engineNEWAPI 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. 16705
    Prev
    Next
    INDEX
    Explanations

    references to foods and ingredients, particularly those related to goat products or goat meat

    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
    274
    +0.14
    0.7%
    377
    +0.13
    0.6%
    199
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.14
    0.03
    326
    +0.13
    0.02
    659
    +0.11
    0.02
    Negative Logits
    eering
    -0.92
    ept
    -0.77
    icator
    -0.77
    ership
    -0.76
    icators
    -0.75
    ications
    -0.73
    hips
    -0.72
    lines
    -0.70
     replay
    -0.70
    ially
    -0.70
    POSITIVE LOGITS
     Cheese
    0.82
    veyard
    0.75
    ificial
    0.74
    keepers
    0.73
     cheese
    0.72
    hedral
    0.72
    hene
    0.71
    cule
    0.71
    WAYS
    0.67
    keeper
    0.67
    Activations Density 0.066%

    No Known Activations