© 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. 37205
    Prev
    Next
    INDEX
    Explanations

    instances of conditional or prohibitive statements

    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
    358
    +0.11
    0.5%
    287
    +0.11
    0.5%
    179
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    287
    +0.11
    0.03
    696
    +0.11
    0.03
    267
    +0.10
    0.03
    Negative Logits
    */
    -0.70
    س
    -0.63
     srfAttach
    -0.62
     thereof
    -0.62
    itud
    -0.60
     attRot
    -0.59
    ibles
    -0.58
    âĿ
    -0.58
    "}],"
    -0.57
    ÙIJ
    -0.57
    POSITIVE LOGITS
    cknowled
    0.88
     Started
    0.81
     Own
    0.75
    itialized
    0.71
    dating
    0.70
    nea
    0.65
    gdala
    0.63
    iltr
    0.62
     Tradable
    0.61
     Fired
    0.61
    Activations Density 0.143%

    No Known Activations