© 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. 14550
    Prev
    Next
    INDEX
    Explanations

    organizations, affiliations, and locations expressed in the context of professional accomplishments

    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
    642
    +0.12
    0.6%
    756
    +0.11
    0.5%
    444
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.12
    0.02
    701
    +0.11
    0.02
    444
    +0.10
    0.02
    Negative Logits
    ancies
    -0.84
    ted
    -0.79
    APE
    -0.65
     shortest
    -0.64
    short
    -0.63
     Asheville
    -0.62
    rons
    -0.61
     hazards
    -0.61
     owl
    -0.60
    ards
    -0.60
    POSITIVE LOGITS
    estro
    1.12
    eus
    1.05
    ñ
    1.05
    ples
    1.04
    pl
    1.02
    ña
    0.97
    ppa
    0.97
    pling
    0.93
    qua
    0.91
    pler
    0.90
    Activations Density 0.021%

    No Known Activations