© 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. 12893
    Prev
    Next
    INDEX
    Explanations

    details related to criminal cases or investigations involving violence

    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
    393
    +0.10
    0.4%
    338
    +0.09
    0.4%
    481
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    43
    +0.10
    0.05
    100
    +0.09
    0.06
    297
    +0.08
    0.05
    Negative Logits
    Ľ
    -0.72
    OND
    -0.67
    -0.67
    atile
    -0.67
    iasis
    -0.66
    orrect
    -0.66
    GET
    -0.64
    sid
    -0.64
    оÐ
    -0.63
    reb
    -0.63
    POSITIVE LOGITS
     etc
    1.26
    etc
    0.89
    whatever
    0.86
     blah
    0.75
     respectively
    0.75
     yeah
    0.72
     rinse
    0.72
     somew
    0.72
     uh
    0.68
     whatever
    0.65
    Activations Density 0.299%

    No Known Activations