© 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. 10132
    Prev
    Next
    INDEX
    Explanations

    the name "Capuçon"

    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
    336
    +0.12
    0.5%
    659
    +0.12
    0.5%
    284
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    284
    +0.12
    0.04
    659
    +0.12
    0.04
    343
    +0.11
    0.03
    Negative Logits
    =-=-
    -0.74
     Payton
    -0.71
     horizont
    -0.71
     Attribution
    -0.67
     skelet
    -0.67
     SERV
    -0.61
     therap
    -0.61
     Cosponsors
    -0.61
    20439
    -0.61
    */(
    -0.60
    POSITIVE LOGITS
    cca
    1.35
    pport
    1.25
    ccess
    1.13
    seless
    1.11
    pta
    1.10
    cius
    1.09
    itton
    1.02
    isine
    1.02
    ña
    1.01
    cci
    0.98
    Activations Density 0.086%

    No Known Activations