© 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. 37215
    Prev
    Next
    INDEX
    Explanations

    the word "on" 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
    471
    +0.12
    0.5%
    135
    +0.10
    0.5%
    293
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    471
    +0.12
    0.04
    135
    +0.10
    0.04
    90
    +0.09
    0.03
    Negative Logits
    soDeliveryDate
    -0.87
    ONSORED
    -0.79
    çĶ°
    -0.75
    rar
    -0.71
    hips
    -0.70
    uum
    -0.70
    bats
    -0.69
    expensive
    -0.68
    NESS
    -0.68
    mask
    -0.68
    POSITIVE LOGITS
     behalf
    1.12
     arrival
    1.07
     occasion
    1.06
     purpose
    0.93
    eday
    0.93
     cue
    0.92
    erous
    0.91
    etime
    0.91
     Sundays
    0.88
    shore
    0.87
    Activations Density 0.093%

    No Known Activations