© 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. 645
    Prev
    Next
    INDEX
    Explanations

    policy-related statements regarding economics and environmental 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
    481
    +0.17
    0.8%
    393
    +0.12
    0.5%
    756
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    225
    +0.17
    0.07
    484
    +0.12
    0.06
    393
    +0.10
    0.07
    Negative Logits
    wic
    -0.95
    DragonMagazine
    -0.84
     Cosponsors
    -0.73
    married
    -0.72
    quished
    -0.72
    essor
    -0.70
    oteric
    -0.69
    ttp
    -0.69
     Bastard
    -0.68
    lication
    -0.68
    POSITIVE LOGITS
     thereby
    1.15
     consequently
    1.04
     diminish
    1.01
     reduce
    1.01
     enhance
    1.00
     hence
    1.00
     promotes
    1.00
     stimulate
    0.97
     prevents
    0.97
     undermine
    0.96
    Activations Density 0.372%

    No Known Activations