© 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 12288
    9. 8-RES_FS12288-JB
    11. 10377
    Prev
    Next
    INDEX
    Explanations

    proper nouns and names of individuals

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    481
    +0.22
    1.1%
    447
    +0.20
    0.9%
    134
    +0.13
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    481
    +0.22
    0.11
    288
    +0.20
    0.12
    401
    +0.13
    0.10
    Negative Logits
     Invention
    -0.66
    olulu
    -0.65
    };
    -0.62
    Anonymous
    -0.61
    !.
    -0.61
    Marginal
    -0.59
    rawdownloadcloneembedreportprint
    -0.57
    idav
    -0.56
    omatic
    -0.56
    foundland
    -0.56
    POSITIVE LOGITS
     should
    1.19
     could
    1.02
     shouldn
    0.99
     ought
    0.98
     lacked
    0.96
     would
    0.95
     lacks
    0.94
     SHOULD
    0.94
     hadn
    0.93
     violated
    0.93
    Activations Density 1.242%

    No Known Activations