© 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. 37182
    Prev
    Next
    INDEX
    Explanations

    instances of the word "away."

    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
    383
    +0.10
    0.5%
    119
    +0.10
    0.4%
    764
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    119
    +0.10
    0.01
    568
    +0.10
    0.01
    764
    +0.10
    0.01
    Negative Logits
     Briggs
    -0.68
     plurality
    -0.64
    umn
    -0.64
    chini
    -0.64
    ingham
    -0.62
    accompan
    -0.61
    million
    -0.60
     chang
    -0.60
    tone
    -0.60
     haircut
    -0.59
    POSITIVE LOGITS
     from
    0.79
     altogether
    0.75
    agy
    0.74
     entirely
    0.71
    ãĤĵ
    0.71
    sites
    0.71
     FROM
    0.70
    Favorite
    0.67
     lest
    0.67
    zx
    0.65
    Activations Density 0.011%

    No Known Activations