© 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. 5234
    Prev
    Next
    INDEX
    Explanations

    proper nouns or names, particularly related to individuals or organizations

    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
    145
    +0.14
    0.6%
    549
    +0.13
    0.6%
    503
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    145
    +0.14
    0.03
    549
    +0.13
    0.02
    741
    +0.11
    0.02
    Negative Logits
    ¬¼
    -0.78
    ĸļ
    -0.73
    eners
    -0.69
    stakes
    -0.63
     compounded
    -0.62
     goodbye
    -0.62
     birth
    -0.61
     functioning
    -0.61
     sterling
    -0.61
     handc
    -0.61
    POSITIVE LOGITS
    ansas
    1.19
    ithmetic
    1.12
    thritis
    1.10
    beit
    1.10
    rang
    1.04
    rington
    1.01
    agon
    1.01
    lington
    1.01
    leigh
    1.01
    rival
    1.00
    Activations Density 0.012%

    No Known Activations