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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 26586
    Prev
    Next
    INDEX
    Explanations

    concepts and terminology related to economics and business processes

    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
    281
    +0.12
    0.5%
    741
    +0.11
    0.5%
    248
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.12
    0.06
    281
    +0.11
    0.04
    611
    +0.11
    0.04
    Negative Logits
    ãĤ±
    -0.84
    outhern
    -0.83
    cale
    -0.82
    yip
    -0.77
    ighthouse
    -0.76
    icum
    -0.75
    lio
    -0.74
    boro
    -0.74
    icable
    -0.73
    elcome
    -0.73
    POSITIVE LOGITS
     holders
    1.05
     rentals
    0.95
     closures
    0.90
    eers
    0.89
     replacements
    0.88
     registrations
    0.88
     modifications
    0.86
     syndrome
    0.85
     pamph
    0.85
     ultras
    0.85
    Activations Density 0.203%

    No Known Activations