© 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. 20874
    Prev
    Next
    INDEX
    Explanations

    expressions related to discussing potential actions and consequences

    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
    720
    +0.12
    0.5%
    172
    +0.11
    0.5%
    268
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.12
    0.05
    172
    +0.11
    0.05
    546
    +0.10
    0.04
    Negative Logits
    MQ
    -0.76
    nice
    -0.71
     mentions
    -0.70
    Favorite
    -0.69
    stad
    -0.69
    Score
    -0.68
    tif
    -0.66
    fit
    -0.66
    pring
    -0.64
     Collider
    -0.63
    POSITIVE LOGITS
     irreversible
    1.27
     contam
    1.14
     destabil
    1.14
     damaging
    1.13
     jeopard
    1.09
     havoc
    1.07
     disastrous
    1.06
     ruin
    1.03
     unintended
    1.03
     catastrophic
    1.01
    Activations Density 0.415%

    No Known Activations