© 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 12288
    9. 8-RES_FS12288-JB
    11. 1762
    Prev
    Next
    INDEX
    Explanations

    details related to crime scenes and criminal investigations

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    694
    +0.10
    0.5%
    373
    +0.10
    0.5%
    752
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    260
    +0.10
    0.09
    694
    +0.10
    0.05
    333
    +0.09
    0.05
    Negative Logits
     Paradox
    -0.70
    ahime
    -0.68
     univers
    -0.60
     wisely
    -0.58
    ularity
    -0.58
     uniqueness
    -0.58
    Quote
    -0.58
     domains
    -0.57
    inctions
    -0.57
     Reply
    -0.56
    POSITIVE LOGITS
     curfew
    0.81
    romptu
    0.78
     gunfire
    0.76
     makeshift
    0.74
     evacuation
    0.72
    ooters
    0.71
     ambulance
    0.70
     patrol
    0.64
     onlook
    0.64
     explosives
    0.63
    Activations Density 1.742%

    No Known Activations