© 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. 18882
    Prev
    Next
    INDEX
    Explanations

    new or unexpected elements or developments

    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
    490
    +0.10
    0.4%
    13
    +0.10
    0.4%
    447
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    540
    +0.10
    0.06
    195
    +0.10
    0.03
    408
    +0.10
    0.06
    Negative Logits
    hovah
    -0.70
    azeera
    -0.69
    inventoryQuantity
    -0.68
    arus
    -0.68
    hner
    -0.68
     etc
    -0.67
    arer
    -0.66
     Unloaded
    -0.64
    heid
    -0.63
    apeake
    -0.63
    POSITIVE LOGITS
    :
    0.85
    :#
    0.83
    :,
    0.80
     namely
    0.80
    !:
    0.79
    :]
    0.78
    ]:
    0.78
     whereby
    0.75
    *:
    0.75
    :(
    0.72
    Activations Density 0.339%

    No Known Activations