© 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. 10243
    Prev
    Next
    INDEX
    Explanations

    phrases related to advocacy and social issues

    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
    139
    +0.14
    0.6%
    697
    +0.11
    0.5%
    496
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    354
    +0.14
    0.11
    697
    +0.11
    0.09
    139
    +0.11
    0.09
    Negative Logits
    CAP
    -0.70
     Upper
    -0.65
    lining
    -0.64
    earch
    -0.62
     Eleven
    -0.61
    itol
    -0.60
    Enlarge
    -0.60
    GMT
    -0.60
    ormal
    -0.56
    UPDATE
    -0.56
    POSITIVE LOGITS
    'll
    1.05
    've
    1.03
    're
    0.97
     cannot
    0.93
     hope
    0.87
     don
    0.85
     know
    0.85
    'd
    0.83
     owe
    0.82
     can
    0.82
    Activations Density 0.504%

    No Known Activations