© 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. 37160
    Prev
    Next
    INDEX
    Explanations

    instances of the conjunction "and" 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
    521
    +0.11
    0.5%
    465
    +0.10
    0.4%
    597
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    350
    +0.11
    0.03
    363
    +0.10
    0.03
    716
    +0.09
    0.03
    Negative Logits
    ibly
    -0.82
    nown
    -0.78
    ONSORED
    -0.77
    ious
    -0.75
    perture
    -0.74
     Petr
    -0.69
    Asset
    -0.69
    icity
    -0.68
    uously
    -0.68
    )].
    -0.67
    POSITIVE LOGITS
     princess
    0.82
     goddess
    0.75
     queen
    0.75
     daughters
    0.69
     limb
    0.67
    orest
    0.66
     sisters
    0.65
    bern
    0.64
    EStream
    0.64
    romeda
    0.63
    Activations Density 0.083%

    No Known Activations