© 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. 5574
    Prev
    Next
    INDEX
    Explanations

    information related to a criminal case or an incident involving law enforcement

    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
    762
    +0.10
    0.5%
    214
    +0.10
    0.4%
    40
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    762
    +0.10
    0.08
    570
    +0.10
    0.01
    728
    +0.10
    0.03
    Negative Logits
    ãĤ´ãĥ³
    -0.81
    Universal
    -0.73
     Expansion
    -0.72
    izons
    -0.72
    introdu
    -0.71
     revamped
    -0.70
    bitious
    -0.69
    translation
    -0.69
    optim
    -0.69
     Quartz
    -0.69
    POSITIVE LOGITS
     Defendant
    1.24
     criminality
    1.19
     manslaughter
    1.16
     defendant
    1.15
     innocent
    1.13
     criminal
    1.11
     unlaw
    1.10
     unlawfully
    1.10
     unlawful
    1.10
     altercation
    1.05
    Activations Density 0.812%

    No Known Activations