© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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 49152
    9. 8-RES_FS49152-JB
    11. 45683
    Prev
    Next
    INDEX
    Explanations

    references to legal charges, specifically misdemeanors and felonies

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    288
    +0.11
    0.5%
    444
    +0.10
    0.4%
    408
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    628
    +0.11
    0.01
    623
    +0.10
    0.01
    155
    +0.09
    0.01
    Negative Logits
    atorial
    -0.94
    reen
    -0.89
    efficients
    -0.84
    icians
    -0.83
    adle
    -0.81
    efficient
    -0.80
    igree
    -0.79
    choes
    -0.78
    obyl
    -0.77
    arya
    -0.77
    POSITIVE LOGITS
     punishable
    1.24
     offenses
    1.15
     misdemeanor
    1.12
     manslaughter
    1.06
     arrest
    0.99
     charges
    0.97
     offense
    0.96
     arrests
    0.95
     misdem
    0.95
     citation
    0.94
    Activations Density 0.009%

    No Known Activations