© 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. 4082
    Prev
    Next
    INDEX
    Explanations

    references to a specific comedian named Ricky Gervais

    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
    554
    +0.11
    0.5%
    277
    +0.10
    0.5%
    497
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.11
    0.01
    588
    +0.10
    0.01
    4
    +0.09
    0.01
    Negative Logits
    DIV
    -0.62
    fine
    -0.61
    ¦
    -0.60
     abund
    -0.60
    ¿½
    -0.59
    Fine
    -0.59
    colour
    -0.59
    fixes
    -0.58
    print
    -0.58
    DX
    -0.56
    POSITIVE LOGITS
    asive
    1.20
    oir
    1.18
    atism
    1.09
    oice
    1.09
    iew
    1.08
    ice
    1.01
    untled
    0.99
    icious
    0.97
    okes
    0.96
    ille
    0.96
    Activations Density 0.014%

    No Known Activations