© 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 768
    9. 8-RES_FS768-JB
    11. 736
    Prev
    Next
    INDEX
    Explanations

    information related to political systems and societal implications, such as voter disenfranchisement, automation impact on jobs, explanations of economic concepts, and arguments related to the Electoral College

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    640
    +0.18
    0.9%
    679
    +0.16
    0.8%
    255
    +0.15
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.18
    0.35
    679
    +0.16
    0.28
    255
    +0.15
    0.24
    Negative Logits
     disemb
    -0.80
     hatch
    -0.71
     aven
    -0.71
     planned
    -0.71
     pocket
    -0.71
     purse
    -0.69
     haul
    -0.69
     exclusively
    -0.69
     unre
    -0.69
     affili
    -0.69
    POSITIVE LOGITS
     Conversely
    1.82
     Therefore
    1.81
     Moreover
    1.79
     Consequently
    1.79
     Furthermore
    1.77
     Likewise
    1.76
     Hence
    1.76
     Similarly
    1.76
     Thus
    1.71
     Additionally
    1.69
    Activations Density 3.576%

    No Known Activations