© 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. 45516
    Prev
    Next
    INDEX
    Explanations

    past tense verbs that imply an action or decision

    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
    373
    +0.12
    0.5%
    446
    +0.10
    0.5%
    54
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.12
    0.04
    715
    +0.10
    0.03
    242
    +0.10
    0.03
    Negative Logits
    ById
    -0.68
    ateg
    -0.67
    gart
    -0.66
    mong
    -0.66
    usat
    -0.64
    WER
    -0.63
    cknowled
    -0.62
    vell
    -0.62
     Cosponsors
    -0.61
    HCR
    -0.61
    POSITIVE LOGITS
     him
    0.68
     them
    0.62
     Lanc
    0.61
     Hitman
    0.61
     Wiz
    0.60
     Stamford
    0.59
     THEM
    0.58
    join
    0.58
    ],"
    0.57
     Velvet
    0.57
    Activations Density 0.190%

    No Known Activations