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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 18957
    Prev
    Next
    INDEX
    Explanations

    information related to human rights abuses and atrocities, particularly focusing on solitary confinement and violent acts

    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
    560
    +0.12
    0.6%
    545
    +0.11
    0.5%
    742
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    560
    +0.12
    0.03
    639
    +0.11
    0.04
    742
    +0.10
    0.04
    Negative Logits
    inventoryQuantity
    -0.92
     hindsight
    -0.81
     Prediction
    -0.78
    soType
    -0.76
     bullish
    -0.76
     Effective
    -0.76
    sense
    -0.75
     commenter
    -0.74
     Seller
    -0.72
     prediction
    -0.72
    POSITIVE LOGITS
     raped
    1.09
     tortured
    1.08
     torture
    1.03
    raped
    1.02
     raping
    0.98
     deprived
    0.98
     forcibly
    0.95
     dehuman
    0.95
     deprivation
    0.95
     interrogated
    0.95
    Activations Density 0.385%

    No Known Activations