© 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 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. 6521
    Prev
    Next
    INDEX
    Explanations

    words related to personal opinions and evaluations

    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
    481
    +0.16
    0.7%
    55
    +0.13
    0.6%
    102
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    55
    +0.16
    0.07
    102
    +0.13
    0.07
    540
    +0.10
    0.05
    Negative Logits
    anwhile
    -1.08
    conservancy
    -0.77
    milo
    -0.76
    pta
    -0.75
    izons
    -0.75
    eday
    -0.73
     roam
    -0.72
    rosis
    -0.71
    afia
    -0.70
    idth
    -0.69
    POSITIVE LOGITS
     incomprehensible
    1.25
     disappointing
    1.24
     ironic
    1.18
     frustrating
    1.17
     confusing
    1.15
     unfair
    1.15
     embarrassing
    1.15
     unsettling
    1.14
     puzzling
    1.14
     ridiculous
    1.13
    Activations Density 0.297%

    No Known Activations