© 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. 9482
    Prev
    Next
    INDEX
    Explanations

    words related to political conflicts and statements of condemnation or disapproval

    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
    400
    +0.13
    0.6%
    262
    +0.11
    0.5%
    29
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    400
    +0.13
    0.05
    262
    +0.11
    0.04
    100
    +0.10
    0.05
    Negative Logits
     queues
    -0.76
    azines
    -0.71
     populate
    -0.71
     transitions
    -0.71
    enhagen
    -0.71
    asio
    -0.68
     branching
    -0.66
     stays
    -0.66
     ~/
    -0.65
     climbs
    -0.65
    POSITIVE LOGITS
     disgrace
    1.18
     unacceptable
    1.15
     hypocritical
    1.15
     irresponsible
    1.14
     unethical
    1.12
     coward
    1.11
     hypocrisy
    1.11
     unworthy
    1.08
     disrespectful
    1.07
     unfair
    1.02
    Activations Density 0.921%

    No Known Activations