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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 5349
    Prev
    Next
    INDEX
    Explanations

    political and government-related discussions and actions

    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
    +0.18
    0.9%
    438
    +0.10
    0.5%
    247
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    764
    +0.18
    0.01
    438
    +0.10
    0.01
    247
    +0.09
    0.01
    Negative Logits
    »Ĵ
    -0.70
     fame
    -0.63
    ŃĶ
    -0.62
    orpor
    -0.60
    «ĺ
    -0.57
     abstinence
    -0.57
    eco
    -0.56
     duty
    -0.56
    agles
    -0.55
    ibilities
    -0.55
    POSITIVE LOGITS
     Continued
    0.85
    telling
    0.83
    hart
    0.75
    tenance
    0.73
    stal
    0.73
    line
    0.72
    bold
    0.70
     Transcript
    0.67
     caption
    0.65
     Story
    0.64
    Activations Density 0.003%

    No Known Activations