© 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 49152
    9. 8-RES_FS49152-JB
    11. 41526
    Prev
    Next
    INDEX
    Explanations

    expressions of opinion or commentary on social issues

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    640
    +0.26
    1.2%
    481
    +0.21
    1.0%
    447
    +0.18
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    640
    +0.26
    0.13
    6
    +0.21
    0.12
    107
    +0.18
    0.10
    Negative Logits
     interpreted
    -0.84
     nonexistent
    -0.79
     perce
    -0.78
     suspected
    -0.78
     sensitive
    -0.78
     perceived
    -0.77
     inclusion
    -0.76
     analys
    -0.76
     arbit
    -0.76
     subt
    -0.76
    POSITIVE LOGITS
    Advertisements
    1.48
    END
    1.40
    Conclusion
    1.38
    ********************************
    1.36
    ________________________________
    1.36
    ________________________
    1.34
    Thank
    1.28
    EDIT
    1.27
    References
    1.25
    Thanks
    1.25
    Activations Density 0.355%

    No Known Activations