© 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 24576
    9. 8-RES_FS24576-JB
    11. 7365
    Prev
    Next
    INDEX
    Explanations

    business and political leaders or professionals

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    447
    +0.19
    0.9%
    524
    +0.11
    0.5%
    679
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    224
    +0.19
    0.07
    258
    +0.11
    0.07
    146
    +0.11
    0.06
    Negative Logits
    inventoryQuantity
    -0.70
    yss
    -0.69
    Owner
    -0.63
    Pg
    -0.61
    ogenesis
    -0.59
    omy
    -0.58
     conquest
    -0.58
    oooo
    -0.57
     gravity
    -0.57
     imposition
    -0.57
    POSITIVE LOGITS
     who
    1.11
     whom
    1.07
    hips
    0.97
     alike
    0.94
    folk
    0.92
     whose
    0.88
     gathered
    0.87
    who
    0.83
     assembled
    0.83
     vying
    0.83
    Activations Density 0.293%

    No Known Activations