© 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. 9864
    Prev
    Next
    INDEX
    Explanations

    text related to curiosity or intrigue

    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
    195
    +0.16
    0.7%
    160
    +0.12
    0.6%
    271
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    271
    +0.16
    0.03
    195
    +0.12
    0.02
    151
    +0.10
    0.03
    Negative Logits
    uts
    -0.77
    uter
    -0.72
    aper
    -0.72
    required
    -0.70
    oise
    -0.69
    avers
    -0.68
    otent
    -0.68
    reditation
    -0.68
    bey
    -0.66
     certify
    -0.66
    POSITIVE LOGITS
     tid
    0.96
     Flavoring
    0.96
     trivia
    0.86
     anecdotes
    0.86
     insights
    0.84
    arios
    0.82
     twists
    0.82
     juxtap
    0.81
     coincidence
    0.79
     sidel
    0.78
    Activations Density 0.103%

    No Known Activations