© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 18691
    Prev
    Next
    INDEX
    Explanations

    people speaking anonymously about a situation with a sense of authority

    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
    373
    +0.10
    0.5%
    412
    +0.10
    0.5%
    760
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    760
    +0.10
    0.04
    410
    +0.10
    0.04
    148
    +0.10
    0.04
    Negative Logits
    joice
    -0.69
     pedest
    -0.64
    tumblr
    -0.63
     fundamentals
    -0.61
    setting
    -0.61
     Tend
    -0.61
     sights
    -0.59
    ãĢIJ
    -0.59
     "$:/
    -0.57
     cataly
    -0.56
    POSITIVE LOGITS
     anonymity
    1.22
     surn
    0.94
     confidentiality
    0.84
     interviewed
    0.82
     pseudonym
    0.81
     anonym
    0.79
    fax
    0.78
    ispers
    0.78
     anonymously
    0.77
     newsp
    0.77
    Activations Density 0.318%

    No Known Activations