© 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. 18154
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal matters and government actions

    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
    481
    +0.15
    0.7%
    665
    +0.12
    0.6%
    282
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.15
    0.08
    608
    +0.12
    0.07
    282
    +0.09
    0.05
    Negative Logits
    borg
    -0.70
    stars
    -0.69
     Tent
    -0.67
     resides
    -0.67
     âĵĺ
    -0.66
    tis
    -0.65
     Verse
    -0.64
    Stars
    -0.64
     Functions
    -0.62
     Healer
    -0.62
    POSITIVE LOGITS
     publicity
    1.01
     morale
    0.97
     awareness
    0.87
     deterrence
    0.82
     embarrassment
    0.80
     confusion
    0.78
     retribution
    0.78
     advertis
    0.77
     exha
    0.77
     appease
    0.77
    Activations Density 0.485%

    No Known Activations