© 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 24576
    9. 8-RES_FS24576-JB
    11. 22166
    Prev
    Next
    INDEX
    Explanations

    terms related to correction and resolution of issues or injustices

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    phrases related to resolving issues or correcting injustices

    oai_token-act-pair · gpt-4o-miniTriggered 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
    405
    +0.13
    0.6%
    608
    +0.10
    0.4%
    559
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    405
    +0.13
    0.05
    507
    +0.10
    0.05
    665
    +0.09
    0.04
    Negative Logits
    ograms
    -0.73
    soft
    -0.68
    etry
    -0.67
    ertodd
    -0.66
    rogens
    -0.66
    google
    -0.63
    ittal
    -0.63
    chie
    -0.63
    acs
    -0.62
    alian
    -0.62
    POSITIVE LOGITS
     grievances
    1.13
     deficiencies
    1.01
    balance
    0.99
     misconceptions
    0.98
     tensions
    0.97
     conflicts
    0.94
     misunderstand
    0.93
     disputes
    0.91
     ambig
    0.90
     disagreements
    0.88
    Activations Density 0.148%

    No Known Activations