© 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 3072
    9. 8-RES_FS3072-JB
    11. 117
    Prev
    Next
    INDEX
    Explanations

    text related to data fusion and analysis, specifically focusing on combining psychological data with location data for predicting future psychological states and behaviors

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    608
    +0.13
    0.6%
    362
    +0.13
    0.6%
    201
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.13
    0.14
    362
    +0.13
    0.12
    669
    +0.09
    0.07
    Negative Logits
    chat
    -0.75
    vier
    -0.72
    eno
    -0.71
    raq
    -0.69
    cussion
    -0.68
     fooled
    -0.67
    cture
    -0.67
    ipeg
    -0.67
    tesy
    -0.67
    je
    -0.67
    POSITIVE LOGITS
     ours
    0.84
     neighbouring
    0.80
     others
    0.78
     existing
    0.77
     other
    0.76
     reality
    0.74
     neighboring
    0.74
    rium
    0.71
     another
    0.71
     Catholicism
    0.71
    Activations Density 1.464%

    No Known Activations