© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    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. 41641
    Prev
    Next
    INDEX
    Explanations

    references to systems and structures in various contexts

    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
    169
    +0.10
    0.5%
    208
    +0.09
    0.4%
    745
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.07
    325
    +0.09
    0.05
    698
    +0.09
    0.05
    Negative Logits
    Merit
    -0.75
    20439
    -0.75
    ickets
    -0.71
    Joined
    -0.69
    nants
    -0.67
    osaurs
    -0.66
    erred
    -0.64
     Joined
    -0.63
    advertising
    -0.63
    borgh
    -0.62
    POSITIVE LOGITS
     proverb
    0.83
     analogy
    0.77
     heterogeneity
    0.75
     playbook
    0.74
     specificity
    0.73
     causation
    0.73
     inertia
    0.71
     itself
    0.71
     multiplier
    0.69
     dilemma
    0.68
    Activations Density 0.414%

    No Known Activations