© 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. 18844
    Prev
    Next
    INDEX
    Explanations

    statistical data related to urban crime rates

    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
    570
    +0.12
    0.6%
    331
    +0.10
    0.5%
    218
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    570
    +0.12
    0.03
    153
    +0.10
    0.04
    111
    +0.09
    0.03
    Negative Logits
     Saga
    -0.67
     Samoa
    -0.60
    WAY
    -0.60
     GSL
    -0.55
    Virtual
    -0.54
    lihood
    -0.54
    :{
    -0.52
     cables
    -0.52
     microphones
    -0.51
    URI
    -0.51
    POSITIVE LOGITS
    pperc
    0.61
    apple
    0.61
    cend
    0.61
    ynes
    0.61
    erenn
    0.59
    til
    0.59
    ]."
    0.58
    chid
    0.58
    imens
    0.57
    nova
    0.56
    Activations Density 0.181%

    No Known Activations