Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 768
    5. 8-RES_FS768-JB
    6. 41
    Prev
    Next
    INDEX
    Explanations

    discussion of specific individuals or entities within a community or organization, likely in a professional or work-related context

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    314
    +0.13
    0.6%
    640
    +0.10
    0.5%
    727
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    727
    +0.13
    0.10
    176
    +0.10
    0.06
    273
    +0.10
    0.12
    Negative Logits
    .","
    -0.79
     thereto
    -0.78
    -"
    -0.72
     thereof
    -0.72
    itiz
    -0.71
    Ïī
    -0.68
     â̦"
    -0.68
     ..."
    -0.67
     ("
    -0.66
    rouse
    -0.65
    POSITIVE LOGITS
    resa
    1.52
    odore
    1.24
    xiety
    1.18
    alyst
    1.05
    withstanding
    1.02
    mosp
    1.00
    swers
    0.98
     Facts
    0.96
    bidden
    0.92
    chieve
    0.89
    Activations Density 4.587%

    No Known Activations